aid stringlengths 9 15 | mid stringlengths 7 10 | abstract stringlengths 78 2.56k | related_work stringlengths 92 1.77k | ref_abstract dict |
|---|---|---|---|---|
1705.06947 | 2952799392 | As the number and the diversity of news outlets on the Web grow, so does the opportunity for "alternative" sources of information to emerge. Using large social networks like Twitter and Facebook, misleading, false, or agenda-driven information can quickly and seamlessly spread online, deceiving people or influencing their opinions. Also, the increased engagement of tightly knit communities, such as Reddit and 4chan, further compounds the problem, as their users initiate and propagate alternative information, not only within their own communities, but also to different ones as well as various social media. In fact, these platforms have become an important piece of the modern information ecosystem, which, thus far, has not been studied as a whole. In this paper, we begin to fill this gap by studying mainstream and alternative news shared on Twitter, Reddit, and 4chan. By analyzing millions of posts around several axes, we measure how mainstream and alternative news flows between these platforms. Our results indicate that alt-right communities within 4chan and Reddit can have a surprising level of influence on Twitter, providing evidence that "fringe" communities often succeed in spreading alternative news to mainstream social networks and the greater Web. | @cite_1 study the notion of competing campaigns in a social network and address the problem of influence limitation to counteract the effect of misinformation. @cite_11 look for the @math users that are most suspected to have originated false information, using a reverse diffusion process along with a ranking process. @cite_29 aim to identify the source of rumors in online social networks by injecting monitoring nodes across the social graph. They propose an algorithm that observes the information received by the monitoring nodes in order to identify the source. They indicate that with sufficient number of monitoring sources they can recognize the source with high accuracy. | {
"cite_N": [
"@cite_29",
"@cite_1",
"@cite_11"
],
"mid": [
"2031490232",
"",
"2024814977"
],
"abstract": [
"Information that propagates through social networks can carry a lot of false claims. For example, rumors on certain topics can propagate rapidly leading to a large number of nodes reporting the same (incorrect) observations. In this paper, we describe an approach for nding the rumor source and assessing the likelihood that a piece of information is in fact a rumor, in the absence of data provenance information. We model the social network as a directed graph, where vertices represent individuals and directed edges represent information ow (e.g., who follows whom on Twitter). A number of monitor nodes are injected into the network whose job is to report data they receive. Our algorithm identies rumors and their sources by observing which of the monitors received the given piece of information and which did not. We show that, with a sucient number of monitor nodes, it is possible to recognize most rumors and their sources with high accuracy.",
"",
"Online Social Networks (OSNs) have recently emerged as one of the most effective channels for information sharing and discovery due to their ability of allowing users to read and create new content simultaneously. While this advantage provides users more rooms to decide which content to follow, it also makes OSNs fertile grounds for the wide spread of misinformation which can lead to undesirable consequences. In order to guarantee the trustworthiness of content sharing in OSNs, it is thus essential to have a strategic investigation on the first and foremost concern: the sources of misinformation. In this paper, we study k-Suspector problem which aims to identify the top k most suspected sources of misinformation. We propose two effective approaches namely ranking-based and optimization-based algorithms. We further extend our solutions to cope with the incompleteness of collected data as well as multiple attacks, which mostly occur in reality. Experimental results on real-world datasets show that our approaches achieve competitive detection ratios in a timely manner in comparison with available methods."
]
} |
1705.07272 | 2617359015 | This paper focuses on real-time all-frequency image-based rendering using an innovative solution for run-time computation of light transport. The approach is based on new results derived for non-linear phase shifting in the Haar wavelet domain. Although image-based methods for real-time rendering of dynamic glossy objects have been proposed, they do not truly scale to all possible frequencies and high sampling rates without trading storage, glossiness, or computational time, while varying both lighting and viewpoint. This is due to the fact that current approaches are limited to precomputed radiance transfer (PRT), which is prohibitively expensive in terms of memory requirements and real-time rendering when both varying light and viewpoint changes are required together with high sampling rates for high frequency lighting of glossy material. On the other hand, current methods cannot handle object rotation, which is one of the paramount issues for all PRT methods using wavelets. This latter problem arises because the precomputed data are defined in a global coordinate system and encoded in the wavelet domain, while the object is rotated in a local coordinate system. At the root of all the above problems is the lack of efficient run-time solution to the nontrivial problem of rotating wavelets (a non-linear phase-shift), which we solve in this paper. | One of the oldest and most straightforward approaches for solving the light integral equation is to approximate the solution using the Monte Carlo method @cite_173 @cite_87 @cite_85 . Global illumination algorithms based on Monte Carlo are general enough to allow estimation of the integral for any material type and light distribution. | {
"cite_N": [
"@cite_85",
"@cite_173",
"@cite_87"
],
"mid": [
"",
"2014208555",
"2138624212"
],
"abstract": [
"",
"1 The general nature of Monte Carlo methods.- 2 Short resume of statistical terms.- 3 Random, pseudorandom, and quasirandom numbers.- 4 Direct simulation.- 5 General principles of the Monte Carlo method.- 6 Conditional Monte Carlo.- 7 Solution of linear operator equations.- 8 Radiation shielding and reactor criticality.- 9 Problems in statistical mechanics.- 10 Long polymer molecules.- 11 Percolation processes.- 12 Multivariable problems.- References.",
"We present an integral equation which generalizes a variety of known rendering algorithms. In the course of discussing a monte carlo solution we also present a new form of variance reduction, called Hierarchical sampling and give a number of elaborations shows that it may be an efficient new technique for a wide variety of monte carlo procedures. The resulting rendering algorithm extends the range of optical phenomena which can be effectively simulated."
]
} |
1705.07272 | 2617359015 | This paper focuses on real-time all-frequency image-based rendering using an innovative solution for run-time computation of light transport. The approach is based on new results derived for non-linear phase shifting in the Haar wavelet domain. Although image-based methods for real-time rendering of dynamic glossy objects have been proposed, they do not truly scale to all possible frequencies and high sampling rates without trading storage, glossiness, or computational time, while varying both lighting and viewpoint. This is due to the fact that current approaches are limited to precomputed radiance transfer (PRT), which is prohibitively expensive in terms of memory requirements and real-time rendering when both varying light and viewpoint changes are required together with high sampling rates for high frequency lighting of glossy material. On the other hand, current methods cannot handle object rotation, which is one of the paramount issues for all PRT methods using wavelets. This latter problem arises because the precomputed data are defined in a global coordinate system and encoded in the wavelet domain, while the object is rotated in a local coordinate system. At the root of all the above problems is the lack of efficient run-time solution to the nontrivial problem of rotating wavelets (a non-linear phase-shift), which we solve in this paper. | Much research has gone into improving Monte Carlo's performance without increasing the number of samples. One of the techniques that has been most effective is importance sampling @cite_18 @cite_84 @cite_171 . Importance sampling relies on sampling mostly in the important'' directions, which is governed by the choice of a sampling distribution function that is similar in shape to the integrand of the function that is being estimated @cite_4 . | {
"cite_N": [
"@cite_18",
"@cite_4",
"@cite_171",
"@cite_84"
],
"mid": [
"1965458840",
"1564611446",
"2043682027",
""
],
"abstract": [
"We introduce structured importance sampling, a new technique for efficiently rendering scenes illuminated by distant natural illumination given in an environment map. Our method handles occlusion, high-frequency lighting, and is significantly faster than alternative methods based on Monte Carlo sampling. We achieve this speedup as a result of several ideas. First, we present a new metric for stratifying and sampling an environment map taking into account both the illumination intensity as well as the expected variance due to occlusion within the scene. We then present a novel hierarchical stratification algorithm that uses our metric to automatically stratify the environment map into regular strata. This approach enables a number of rendering optimizations, such as pre-integrating the illumination within each stratum to eliminate noise at the cost of adding bias, and sorting the strata to reduce the number of sample rays. We have rendered several scenes illuminated by natural lighting, and our results indicate that structured importance sampling is better than the best previous Monte Carlo techniques, requiring one to two orders of magnitude fewer samples for the same image quality.",
"Physically Based Rendering: From Theory to Implementation, Third Edition, describes both the mathematical theory behind a modern photorealistic rendering system and its practical implementation. Through a method known as 'literate programming', the authors combine human-readable documentation and source code into a single reference that is specifically designed to aid comprehension. The result is a stunning achievement in graphics education. Through the ideas and software in this book, users will learn to design and employ a fully-featured rendering system for creating stunning imagery. This completely updated and revised edition includes new coverage on ray-tracing hair and curves primitives, numerical precision issues with ray tracing, LBVHs, realistic camera models, the measurement equation, and much more. It is a must-have, full color resource on physically-based rendering. Presents up-to-date revisions of the seminal reference on rendering, including new sections on bidirectional path tracing, numerical robustness issues in ray tracing, realistic camera models, and subsurface scattering Provides the source code fora complete rendering systemallowing readers to get up and running fast Includes a unique indexing feature, literate programming, that lists the locations of each function, variable, and method on the page where they are first describedServes as an essential resource on physically-based rendering",
"This paper presents a novel method for efficiently generating a good sampling pattern given an importance density over a 2D domain. A Penrose tiling is hierarchically subdivided creating a sufficiently large number of sample points. These points are numbered using the Fibonacci number system, and these numbers are used to threshold the samples against the local value of the importance density. Pre-computed correction vectors, obtained using relaxation, are used to improve the spectral characteristics of the sampling pattern. The technique is deterministic and very fast; the sampling time grows linearly with the required number of samples. We illustrate our technique with importance-based environment mapping, but the technique is versatile enough to be used in a large variety of computer graphics applications, such as light transport calculations, digital halftoning, geometry processing, and various rendering techniques.",
""
]
} |
1705.07272 | 2617359015 | This paper focuses on real-time all-frequency image-based rendering using an innovative solution for run-time computation of light transport. The approach is based on new results derived for non-linear phase shifting in the Haar wavelet domain. Although image-based methods for real-time rendering of dynamic glossy objects have been proposed, they do not truly scale to all possible frequencies and high sampling rates without trading storage, glossiness, or computational time, while varying both lighting and viewpoint. This is due to the fact that current approaches are limited to precomputed radiance transfer (PRT), which is prohibitively expensive in terms of memory requirements and real-time rendering when both varying light and viewpoint changes are required together with high sampling rates for high frequency lighting of glossy material. On the other hand, current methods cannot handle object rotation, which is one of the paramount issues for all PRT methods using wavelets. This latter problem arises because the precomputed data are defined in a global coordinate system and encoded in the wavelet domain, while the object is rotated in a local coordinate system. At the root of all the above problems is the lack of efficient run-time solution to the nontrivial problem of rotating wavelets (a non-linear phase-shift), which we solve in this paper. | Imposing constraints to simplify the integral to achieve real-time or interactive rates is a necessity with spherical harmonics, when glossy or general BRDF's are used. One method is to use low sampling rates, which is equivalent to band-limiting the illumination. This approach is used by in @cite_159 , @cite_0 and @cite_52 . However, band-limiting removes high frequency components, which blurs lighting detail. For diffuse materials, the error can be very low @cite_166 . However, this approach is not suitable for glossy materials, which need the high frequency for an efficient representation. Another approach is to reduce the dimensionality of the problem by fixing the light or the viewing direction as in @cite_2 . This approach, however, restricts the dynamics of the scene. | {
"cite_N": [
"@cite_159",
"@cite_52",
"@cite_0",
"@cite_166",
"@cite_2"
],
"mid": [
"2164369380",
"2106598309",
"",
"2105649179",
"2111999986"
],
"abstract": [
"We present a new, real-time method for rendering diffuse and glossy objects in low-frequency lighting environments that captures soft shadows, interreflections, and caustics. As a preprocess, a novel global transport simulator creates functions over the object's surface representing transfer of arbitrary, low-frequency incident lighting into transferred radiance which includes global effects like shadows and interreflections from the object onto itself. At run-time, these transfer functions are applied to actual incident lighting. Dynamic, local lighting is handled by sampling it close to the object every frame; the object can also be rigidly rotated with respect to the lighting and vice versa. Lighting and transfer functions are represented using low-order spherical harmonics. This avoids aliasing and evaluates efficiently on graphics hardware by reducing the shading integral to a dot product of 9 to 25 element vectors for diffuse receivers. Glossy objects are handled using matrices rather than vectors. We further introduce functions for radiance transfer from a dynamic lighting environment through a preprocessed object to neighboring points in space. These allow soft shadows and caustics from rigidly moving objects to be cast onto arbitrary, dynamic receivers. We demonstrate real-time global lighting effects with this approach.",
"We compress storage and accelerate performance of precomputed radiance transfer (PRT), which captures the way an object shadows, scatters, and reflects light. PRT records over many surface points a transfer matrix. At run-time, this matrix transforms a vector of spherical harmonic coefficients representing distant, low-frequency source lighting into exiting radiance. Per-point transfer matrices form a high-dimensional surface signal that we compress using clustered principal component analysis (CPCA), which partitions many samples into fewer clusters each approximating the signal as an affine subspace. CPCA thus reduces the high-dimensional transfer signal to a low-dimensional set of per-point weights on a per-cluster set of representative matrices. Rather than computing a weighted sum of representatives and applying this result to the lighting, we apply the representatives to the lighting per-cluster (on the CPU) and weight these results per-point (on the GPU). Since the output of the matrix is lower-dimensional than the matrix itself, this reduces computation. We also increase the accuracy of encoded radiance functions with a new least-squares optimal projection of spherical harmonics onto the hemisphere. We describe an implementation on graphics hardware that performs real-time rendering of glossy objects with dynamic self-shadowing and interreflection without fixing the view or light as in previous work. Our approach also allows significantly increased lighting frequency when rendering diffuse objects and includes subsurface scattering.",
"",
"We consider the rendering of diffuse objects under distant illumination, as specified by an environment map. Using an analytic expression for the irradiance in terms of spherical harmonic coefficients of the lighting, we show that one needs to compute and use only 9 coefficients, corresponding to the lowest-frequency modes of the illumination, in order to achieve average errors of only 1 . In other words, the irradiance is insensitive to high frequencies in the lighting, and is well approximated using only 9 parameters. In fact, we show that the irradiance can be procedurally represented simply as a quadratic polynomial in the cartesian components of the surface normal, and give explicit formulae. These observations lead to a simple and efficient procedural rendering algorithm amenable to hardware implementation, a prefiltering method up to three orders of magnitude faster than previous techniques, and new representations for lighting design and image-based rendering.",
"Real-time shading using general (e.g., anisotropic) BRDFs has so far been limited to a few point or directional light sources. We extend such shading to smooth, area lighting using a low-order spherical harmonic basis for the lighting environment. We represent the 4D product function of BRDF times the cosine factor (dot product of the incident lighting and surface normal vectors) as a 2D table of spherical harmonic coefficients. Each table entry represents, for a single view direction, the integral of this product function times lighting on the hemisphere expressed in spherical harmonics. This reduces the shading integral to a simple dot product of 25 component vectors, easily evaluatable on PC graphics hardware. Non-trivial BRDF models require rotating the lighting coefficients to a local frame at each point on an object, currently forming the computational bottleneck. Real-time results can be achieved by fixing the view to allow dynamic lighting or vice versa. We also generalize a previous method for precomputed radiance transfer to handle general BRDF shading. This provides shadows and interreflections that respond in real-time to lighting changes on a preprocessed object of arbitrary material (BRDF) type."
]
} |
1705.07272 | 2617359015 | This paper focuses on real-time all-frequency image-based rendering using an innovative solution for run-time computation of light transport. The approach is based on new results derived for non-linear phase shifting in the Haar wavelet domain. Although image-based methods for real-time rendering of dynamic glossy objects have been proposed, they do not truly scale to all possible frequencies and high sampling rates without trading storage, glossiness, or computational time, while varying both lighting and viewpoint. This is due to the fact that current approaches are limited to precomputed radiance transfer (PRT), which is prohibitively expensive in terms of memory requirements and real-time rendering when both varying light and viewpoint changes are required together with high sampling rates for high frequency lighting of glossy material. On the other hand, current methods cannot handle object rotation, which is one of the paramount issues for all PRT methods using wavelets. This latter problem arises because the precomputed data are defined in a global coordinate system and encoded in the wavelet domain, while the object is rotated in a local coordinate system. At the root of all the above problems is the lack of efficient run-time solution to the nontrivial problem of rotating wavelets (a non-linear phase-shift), which we solve in this paper. | @cite_10 used non-linear wavelet approximation to achieve better localization than spherical harmonics and were successful at representing different degrees of shadowing due to wavelets' excellent capability in handling information at different scales. They, however, reduced the dimensionality of the integral to simplify it by fixing the viewing direction. @cite_133 developed a method for solving the triple-product integral using two-dimensional non-separable Haar wavelets to avoid reducing the dimensionality of the problem and the low sampling rate inherent in spherical harmonics methods. to a triple product: | {
"cite_N": [
"@cite_10",
"@cite_133"
],
"mid": [
"2027509884",
"2076669957"
],
"abstract": [
"We present a method, based on pre-computed light transport, for real-time rendering of objects under all-frequency, time-varying illumination represented as a high-resolution environment map. Current techniques are limited to small area lights, with sharp shadows, or large low-frequency lights, with very soft shadows. Our main contribution is to approximate the environment map in a wavelet basis, keeping only the largest terms (this is known as a non-linear approximation). We obtain further compression by encoding the light transport matrix sparsely but accurately in the same basis. Rendering is performed by multiplying a sparse light vector by a sparse transport matrix, which is very fast. For accurate rendering, using non-linear wavelets is an order of magnitude faster than using linear spherical harmonics, the current best technique.",
"This paper focuses on efficient rendering based on pre-computed light transport, with realistic materials and shadows under all-frequency direct lighting such an environment maps. The basic difficulty is representation and computation in the 6D space of light direction, view direction, and surface position. While image-based and synthetic methods for real-time rendering have been proposed, they do not scale to high sampling rates with variation of both lighting and viewpoint. Current approaches are therefore limited to lower dimensionality (only lighting or viewpoint variation, not both) or lower sampling rates (low frequency lighting and materials). We propose a new mathematical and computational analysis of pre-computed light transport. We use factored forms, separately pre-computing and representing visibility and material properties. Rendering then requires computing triple product integrals at each vertex, involving the lighting, visibility and BRDF. Our main contribution is a general analysis of these triple product integrals, which are likely to have broad applicability in computer graphics and numerical analysis. We first determine the computational complexity in a number of bases like point samples, spherical harmonics and wavelets. We then give efficient linear and sublinear-time algorithms for Haar wavelets, incorporating non-linear wavelet approximation of lighting and BRDFs. Practically, we demonstrate rendering of images under new lighting and viewing conditions in a few seconds, significantly faster than previous techniques."
]
} |
1705.07272 | 2617359015 | This paper focuses on real-time all-frequency image-based rendering using an innovative solution for run-time computation of light transport. The approach is based on new results derived for non-linear phase shifting in the Haar wavelet domain. Although image-based methods for real-time rendering of dynamic glossy objects have been proposed, they do not truly scale to all possible frequencies and high sampling rates without trading storage, glossiness, or computational time, while varying both lighting and viewpoint. This is due to the fact that current approaches are limited to precomputed radiance transfer (PRT), which is prohibitively expensive in terms of memory requirements and real-time rendering when both varying light and viewpoint changes are required together with high sampling rates for high frequency lighting of glossy material. On the other hand, current methods cannot handle object rotation, which is one of the paramount issues for all PRT methods using wavelets. This latter problem arises because the precomputed data are defined in a global coordinate system and encoded in the wavelet domain, while the object is rotated in a local coordinate system. At the root of all the above problems is the lack of efficient run-time solution to the nontrivial problem of rotating wavelets (a non-linear phase-shift), which we solve in this paper. | Rhis formulation was general and worked for any basis. However, solving for @math and efficiently computing the triple sum in ) was not a trivial matter. One can refer to @cite_133 for an in-depth analysis of the computational complexity for the different methods to solve the triple sum. | {
"cite_N": [
"@cite_133"
],
"mid": [
"2076669957"
],
"abstract": [
"This paper focuses on efficient rendering based on pre-computed light transport, with realistic materials and shadows under all-frequency direct lighting such an environment maps. The basic difficulty is representation and computation in the 6D space of light direction, view direction, and surface position. While image-based and synthetic methods for real-time rendering have been proposed, they do not scale to high sampling rates with variation of both lighting and viewpoint. Current approaches are therefore limited to lower dimensionality (only lighting or viewpoint variation, not both) or lower sampling rates (low frequency lighting and materials). We propose a new mathematical and computational analysis of pre-computed light transport. We use factored forms, separately pre-computing and representing visibility and material properties. Rendering then requires computing triple product integrals at each vertex, involving the lighting, visibility and BRDF. Our main contribution is a general analysis of these triple product integrals, which are likely to have broad applicability in computer graphics and numerical analysis. We first determine the computational complexity in a number of bases like point samples, spherical harmonics and wavelets. We then give efficient linear and sublinear-time algorithms for Haar wavelets, incorporating non-linear wavelet approximation of lighting and BRDFs. Practically, we demonstrate rendering of images under new lighting and viewing conditions in a few seconds, significantly faster than previous techniques."
]
} |
1705.07272 | 2617359015 | This paper focuses on real-time all-frequency image-based rendering using an innovative solution for run-time computation of light transport. The approach is based on new results derived for non-linear phase shifting in the Haar wavelet domain. Although image-based methods for real-time rendering of dynamic glossy objects have been proposed, they do not truly scale to all possible frequencies and high sampling rates without trading storage, glossiness, or computational time, while varying both lighting and viewpoint. This is due to the fact that current approaches are limited to precomputed radiance transfer (PRT), which is prohibitively expensive in terms of memory requirements and real-time rendering when both varying light and viewpoint changes are required together with high sampling rates for high frequency lighting of glossy material. On the other hand, current methods cannot handle object rotation, which is one of the paramount issues for all PRT methods using wavelets. This latter problem arises because the precomputed data are defined in a global coordinate system and encoded in the wavelet domain, while the object is rotated in a local coordinate system. At the root of all the above problems is the lack of efficient run-time solution to the nontrivial problem of rotating wavelets (a non-linear phase-shift), which we solve in this paper. | Haar wavelets provide an efficient solution for solving the triple sum due to the following simple theorem @cite_133 : The Tripling Coefficient Theorem The integral of three 2D Haar basis functions is non-zero if and only if one of the following three cases hold: All three are scaling functions, that is, @math . All three occupy the same wavelet square and all are different wavelet types, that is @math , where wavelets are at level @math . Two are identical wavelets, and the third is either the scaling function or a wavelet that overlaps at a strictly coarser level, that is, @math , where the third function exists at level @math . | {
"cite_N": [
"@cite_133"
],
"mid": [
"2076669957"
],
"abstract": [
"This paper focuses on efficient rendering based on pre-computed light transport, with realistic materials and shadows under all-frequency direct lighting such an environment maps. The basic difficulty is representation and computation in the 6D space of light direction, view direction, and surface position. While image-based and synthetic methods for real-time rendering have been proposed, they do not scale to high sampling rates with variation of both lighting and viewpoint. Current approaches are therefore limited to lower dimensionality (only lighting or viewpoint variation, not both) or lower sampling rates (low frequency lighting and materials). We propose a new mathematical and computational analysis of pre-computed light transport. We use factored forms, separately pre-computing and representing visibility and material properties. Rendering then requires computing triple product integrals at each vertex, involving the lighting, visibility and BRDF. Our main contribution is a general analysis of these triple product integrals, which are likely to have broad applicability in computer graphics and numerical analysis. We first determine the computational complexity in a number of bases like point samples, spherical harmonics and wavelets. We then give efficient linear and sublinear-time algorithms for Haar wavelets, incorporating non-linear wavelet approximation of lighting and BRDFs. Practically, we demonstrate rendering of images under new lighting and viewing conditions in a few seconds, significantly faster than previous techniques."
]
} |
1705.07272 | 2617359015 | This paper focuses on real-time all-frequency image-based rendering using an innovative solution for run-time computation of light transport. The approach is based on new results derived for non-linear phase shifting in the Haar wavelet domain. Although image-based methods for real-time rendering of dynamic glossy objects have been proposed, they do not truly scale to all possible frequencies and high sampling rates without trading storage, glossiness, or computational time, while varying both lighting and viewpoint. This is due to the fact that current approaches are limited to precomputed radiance transfer (PRT), which is prohibitively expensive in terms of memory requirements and real-time rendering when both varying light and viewpoint changes are required together with high sampling rates for high frequency lighting of glossy material. On the other hand, current methods cannot handle object rotation, which is one of the paramount issues for all PRT methods using wavelets. This latter problem arises because the precomputed data are defined in a global coordinate system and encoded in the wavelet domain, while the object is rotated in a local coordinate system. At the root of all the above problems is the lack of efficient run-time solution to the nontrivial problem of rotating wavelets (a non-linear phase-shift), which we solve in this paper. | Schr " o der and Sweldens @cite_172 constructed biorthogonal wavelets on the sphere using the lifting scheme. Their method achieved better localization than spherical harmonics for high-frequency glossy materials and was defined directly on the sphere. However, it was also not rotation-invariant and did not lend itself to solving the rendering triple integral equation in an efficient manner. | {
"cite_N": [
"@cite_172"
],
"mid": [
"2012909180"
],
"abstract": [
"Wavelets have proven to be powerful bases for use in numerical analysis and signal processing. Their power lies in the fact that they only require a small number of coefficients to represent general functions and large data sets accurately. This allows compression and efficient computations. Classical constructions have been limited to simple domains such as intervals and rectangles. In this paper we present a wavelet construction for scalar functions defined on the sphere. We show how biorthogonal wavelets with custom properties can be constructed with the lifting scheme. The bases are extremely easy to implement and allow fully adaptive subdivisions. We give examples of functions defined on the sphere, such as topographic data, bidirectional reflection distribution functions, and illumination, and show how they can be efficiently represented with spherical wavelets. CR"
]
} |
1705.07272 | 2617359015 | This paper focuses on real-time all-frequency image-based rendering using an innovative solution for run-time computation of light transport. The approach is based on new results derived for non-linear phase shifting in the Haar wavelet domain. Although image-based methods for real-time rendering of dynamic glossy objects have been proposed, they do not truly scale to all possible frequencies and high sampling rates without trading storage, glossiness, or computational time, while varying both lighting and viewpoint. This is due to the fact that current approaches are limited to precomputed radiance transfer (PRT), which is prohibitively expensive in terms of memory requirements and real-time rendering when both varying light and viewpoint changes are required together with high sampling rates for high frequency lighting of glossy material. On the other hand, current methods cannot handle object rotation, which is one of the paramount issues for all PRT methods using wavelets. This latter problem arises because the precomputed data are defined in a global coordinate system and encoded in the wavelet domain, while the object is rotated in a local coordinate system. At the root of all the above problems is the lack of efficient run-time solution to the nontrivial problem of rotating wavelets (a non-linear phase-shift), which we solve in this paper. | @cite_11 parameterized the spherical functions using geometry maps @cite_132 and provided a solution for wavelet rotation using precomputed rotation matrices. They precomputed and stored the rotation matrices and used them to rotate the light coefficients into the local space of the BRDF. However, their solution was brute force rather than analytic, since it would be impossible to derive rotation formulae for the frequency domain with the geometry map representation. | {
"cite_N": [
"@cite_132",
"@cite_11"
],
"mid": [
"2005999035",
"2026173400"
],
"abstract": [
"The traditional approach for parametrizing a surface involves cutting it into charts and mapping these piecewise onto a planar domain. We introduce a robust technique for directly parametrizing a genus-zero surface onto a spherical domain. A key ingredient for making such a parametrization practical is the minimization of a stretch-based measure, to reduce scale-distortion and thereby prevent undersampling. Our second contribution is a scheme for sampling the spherical domain using uniformly subdivided polyhedral domains, namely the tetrahedron, octahedron, and cube. We show that these particular semi-regular samplings can be conveniently represented as completely regular 2D grids, i.e. geometry images. Moreover, these images have simple boundary extension rules that aid many processing operations. Applications include geometry remeshing, level-of-detail, morphing, compression, and smooth surface subdivision.",
"Real-time shading with environment maps requires the ability to rotate the global lighting to each surface point's local coordinate frame. Although extensive previous work has studied rotation of functions represented by spherical harmonics, little work has investigated efficient rotation of wavelets. Wavelets are superior at approximating high frequency signals such as detailed high dynamic range lighting and very shiny BRDFs, but present difficulties for interactive rendering due to the lack of an analytic solution for rotation. In this paper we present an efficient computational solution for wavelet rotation using precomputed matrices. Each matrix represents the linear transformation of source wavelet bases defined in the global coordinate frame to target wavelet bases defined in sampled local frames. Since wavelets have compact support, these matrices are very sparse, enabling efficient storage and fast computation at run-time. In this paper, we focus on the application of our technique to interactive environment map rendering. We show that using these matrices allows us to evaluate the integral of dynamic lighting with dynamic BRDFs at interactive rates, incorporating efficient non-linear approximation of both illumination and reflection. Our technique improves on previous work by eliminating the need for prefiltering environment maps, and is thus significantly faster for accurate rendering of dynamic environment lighting with high frequency reflection effects."
]
} |
1705.07267 | 2618463334 | In this paper, we extend an attention-based neural machine translation (NMT) model by allowing it to access an entire training set of parallel sentence pairs even after training. The proposed approach consists of two stages. In the first stage--retrieval stage--, an off-the-shelf, black-box search engine is used to retrieve a small subset of sentence pairs from a training set given a source sentence. These pairs are further filtered based on a fuzzy matching score based on edit distance. In the second stage--translation stage--, a novel translation model, called translation memory enhanced NMT (TM-NMT), seamlessly uses both the source sentence and a set of retrieved sentence pairs to perform the translation. Empirical evaluation on three language pairs (En-Fr, En-De, and En-Es) shows that the proposed approach significantly outperforms the baseline approach and the improvement is more significant when more relevant sentence pairs were retrieved. | @cite_5 proposed an automatic image caption generation model based on nearest neighbours. In their approach, a given image is queried against a set of training pairs of images and their corresponding captions. They then proposed to use a median caption among those nearest neighboring captions, as a generated caption of the given image. This approach shares some similarity with the first stage of the proposed SEG-NMT. However, unlike their approach, we learn to generate a sentence rather than simply choose one among retrieved ones. | {
"cite_N": [
"@cite_5"
],
"mid": [
"1706899115"
],
"abstract": [
"We explore a variety of nearest neighbor baseline approaches for image captioning. These approaches find a set of nearest neighbor images in the training set from which a caption may be borrowed for the query image. We select a caption for the query image by finding the caption that best represents the \"consensus\" of the set of candidate captions gathered from the nearest neighbor images. When measured by automatic evaluation metrics on the MS COCO caption evaluation server, these approaches perform as well as many recent approaches that generate novel captions. However, human studies show that a method that generates novel captions is still preferred over the nearest neighbor approach."
]
} |
1705.07267 | 2618463334 | In this paper, we extend an attention-based neural machine translation (NMT) model by allowing it to access an entire training set of parallel sentence pairs even after training. The proposed approach consists of two stages. In the first stage--retrieval stage--, an off-the-shelf, black-box search engine is used to retrieve a small subset of sentence pairs from a training set given a source sentence. These pairs are further filtered based on a fuzzy matching score based on edit distance. In the second stage--translation stage--, a novel translation model, called translation memory enhanced NMT (TM-NMT), seamlessly uses both the source sentence and a set of retrieved sentence pairs to perform the translation. Empirical evaluation on three language pairs (En-Fr, En-De, and En-Es) shows that the proposed approach significantly outperforms the baseline approach and the improvement is more significant when more relevant sentence pairs were retrieved. | @cite_2 proposed a memory network for large-scale simple question-answering using an entire Freebase. The output module of the memory network used simple @math -gram matching to create a small set of candidate facts from the Freebase. Each of these candidates was scored by the memory network to create a representation used by the response module. This is similar to our approach in that it exploits a black-box search module ( @math -gram matching) for generating a small candidate set. | {
"cite_N": [
"@cite_2"
],
"mid": [
"580074167"
],
"abstract": [
"Training large-scale question answering systems is complicated because training sources usually cover a small portion of the range of possible questions. This paper studies the impact of multitask and transfer learning for simple question answering; a setting for which the reasoning required to answer is quite easy, as long as one can retrieve the correct evidence given a question, which can be difficult in large-scale conditions. To this end, we introduce a new dataset of 100k questions that we use in conjunction with existing benchmarks. We conduct our study within the framework of Memory Networks (, 2015) because this perspective allows us to eventually scale up to more complex reasoning, and show that Memory Networks can be successfully trained to achieve excellent performance."
]
} |
1705.07267 | 2618463334 | In this paper, we extend an attention-based neural machine translation (NMT) model by allowing it to access an entire training set of parallel sentence pairs even after training. The proposed approach consists of two stages. In the first stage--retrieval stage--, an off-the-shelf, black-box search engine is used to retrieve a small subset of sentence pairs from a training set given a source sentence. These pairs are further filtered based on a fuzzy matching score based on edit distance. In the second stage--translation stage--, a novel translation model, called translation memory enhanced NMT (TM-NMT), seamlessly uses both the source sentence and a set of retrieved sentence pairs to perform the translation. Empirical evaluation on three language pairs (En-Fr, En-De, and En-Es) shows that the proposed approach significantly outperforms the baseline approach and the improvement is more significant when more relevant sentence pairs were retrieved. | A similar approach was very recently proposed for deep reinforcement learning by @cite_1 , where they store pairs of observed state and the corresponding (estimated) value in a key-value memory to build a non-parametric deep Q network. We consider it as a confirmation of the general applicability of the proposed approach to a wider array of problems in machine learning. In the context of neural machine translation, @cite_0 also proposed to use an external key-value memory to remember training examples in the test time. Due to the lack of efficient search mechanism, they do not update the memory jointly with the translation model, unlike the proposed approach in this paper. | {
"cite_N": [
"@cite_0",
"@cite_1"
],
"mid": [
"2953044442",
"2951544594"
],
"abstract": [
"Despite recent advances, memory-augmented deep neural networks are still limited when it comes to life-long and one-shot learning, especially in remembering rare events. We present a large-scale life-long memory module for use in deep learning. The module exploits fast nearest-neighbor algorithms for efficiency and thus scales to large memory sizes. Except for the nearest-neighbor query, the module is fully differentiable and trained end-to-end with no extra supervision. It operates in a life-long manner, i.e., without the need to reset it during training. Our memory module can be easily added to any part of a supervised neural network. To show its versatility we add it to a number of networks, from simple convolutional ones tested on image classification to deep sequence-to-sequence and recurrent-convolutional models. In all cases, the enhanced network gains the ability to remember and do life-long one-shot learning. Our module remembers training examples shown many thousands of steps in the past and it can successfully generalize from them. We set new state-of-the-art for one-shot learning on the Omniglot dataset and demonstrate, for the first time, life-long one-shot learning in recurrent neural networks on a large-scale machine translation task.",
"Deep reinforcement learning methods attain super-human performance in a wide range of environments. Such methods are grossly inefficient, often taking orders of magnitudes more data than humans to achieve reasonable performance. We propose Neural Episodic Control: a deep reinforcement learning agent that is able to rapidly assimilate new experiences and act upon them. Our agent uses a semi-tabular representation of the value function: a buffer of past experience containing slowly changing state representations and rapidly updated estimates of the value function. We show across a wide range of environments that our agent learns significantly faster than other state-of-the-art, general purpose deep reinforcement learning agents."
]
} |
1705.07267 | 2618463334 | In this paper, we extend an attention-based neural machine translation (NMT) model by allowing it to access an entire training set of parallel sentence pairs even after training. The proposed approach consists of two stages. In the first stage--retrieval stage--, an off-the-shelf, black-box search engine is used to retrieve a small subset of sentence pairs from a training set given a source sentence. These pairs are further filtered based on a fuzzy matching score based on edit distance. In the second stage--translation stage--, a novel translation model, called translation memory enhanced NMT (TM-NMT), seamlessly uses both the source sentence and a set of retrieved sentence pairs to perform the translation. Empirical evaluation on three language pairs (En-Fr, En-De, and En-Es) shows that the proposed approach significantly outperforms the baseline approach and the improvement is more significant when more relevant sentence pairs were retrieved. | One important property of the proposed SEG-NMT is that it relies on an external, black-box search engine to retrieve relevant translation pairs. Such a search engine is used both during training and testing, and an obvious next step is to allow the proposed SEG-NMT to more intelligently query the search engine, for instance, by reformulating a given source sentence. Recently, @cite_4 proposed task-oriented query reformulation in which a neural network is trained to use a black-box search engine to maximize the recall of relevant documents, which can be integrated into the proposed SEG-NMT. We leave this as future work. | {
"cite_N": [
"@cite_4"
],
"mid": [
"2607188338"
],
"abstract": [
"Search engines play an important role in our everyday lives by assisting us in finding the information we need. When we input a complex query, however, results are often far from satisfactory. In this work, we introduce a query reformulation system based on a neural network that rewrites a query to maximize the number of relevant documents returned. We train this neural network with reinforcement learning. The actions correspond to selecting terms to build a reformulated query, and the reward is the document recall. We evaluate our approach on three datasets against strong baselines and show a relative improvement of 5-20 in terms of recall. Furthermore, we present a simple method to estimate a conservative upper-bound performance of a model in a particular environment and verify that there is still large room for improvements."
]
} |
1705.06629 | 2615361354 | Relying on ubiquitous Internet connectivity, applications on mobile devices frequently perform web requests during their execution. They fetch data for users to interact with, invoke remote functionalities, or send user-generated content or meta-data. These requests collectively reveal common practices of mobile application development, like what external services are used and how, and they point to possible negative effects like security and privacy violations, or impacts on battery life. In this paper, we assess different ways to analyze what web requests Android applications make. We start by presenting dynamic data collected from running 20 randomly selected Android applications and observing their network activity. Next, we present a static analysis tool, Stringoid, that analyzes string concatenations in Android applications to estimate constructed URL strings. Using Stringoid, we extract URLs from 30, 000 Android applications, and compare the performance with a simpler constant extraction analysis. Finally, we present a discussion of the advantages and limitations of dynamic and static analyses when extracting URLs, as we compare the data extracted by Stringoid from the same 20 applications with the dynamically collected data. | Related work has addressed the problem of understanding the network activities of mobile applications through dynamic analyses. A set of works assesses mobile application network traffic for security and privacy reasons. For example, it attempts to detect malware that infects Android applications based on observing applications' network traffic patterns, relying on features like request sizes, frequencies, and IP addresses @cite_8 . Or, a tool called Securacy, installed on mobile devices, monitors IP addresses and ports invoked from the device, and checks whether secure connections (HTTPS) are used to identify potential security violations @cite_17 . Other works attempt to associate monitored requests with the mobile applications from which they originate, relying for example on HTTP headers @cite_2 or on previously learned network patterns of in-app advertisement services used by applications @cite_16 . All of these works are either limited to un-encrypted traffic, or rely on information that can be collected in any case, like IP addresses. In contrast, in our dynamic analysis, we used a proxy server with a man-in-the-middle facility to obtain information about the URLs being invoked even when traffic is encrypted. | {
"cite_N": [
"@cite_16",
"@cite_2",
"@cite_17",
"@cite_8"
],
"mid": [
"2152047049",
"2002658930",
"2050557001",
"2046843253"
],
"abstract": [
"Recent years have seen an explosive growth in the number of mobile devices such as smart phones and tablets. This has resulted in a growing need of the operators to understand the usage patterns of the mobile apps used on these devices. Previous studies in this area have relied on volunteers using instrumented devices or using fields in the HTTP traffic such as User-Agent to identify the apps in network traces. However, the results of the former approach are difficult to be extrapolated to real-world scenario while the latter approach is not applicable to platforms like Android where developers generally use generic strings, that can not be used to identify the apps, in the User-Agent field. In this paper, we present a novel way of identifying Android apps in network traces using mobile in-app advertisements. Our preliminary experiments with real world traces show that this technique is promising for large scale mobile app usage pattern studies. We also present an analysis of the official Android market place from an advertising perspective.",
"We present SAMPLES: Self Adaptive Mining of Persistent LExical Snippets; a systematic framework for classifying network traffic generated by mobile applications. SAMPLES constructs conjunctive rules, in an automated fashion, through a supervised methodology over a set of labeled flows (the training set). Each conjunctive rule corresponds to the lexical context, associated with an application identifier found in a snippet of the HTTP header, and is defined by: (a) the identifier type, (b) the HTTP header-field it occurs in, and (c) the prefix suffix surrounding its occurrence. Subsequently, these conjunctive rules undergo an aggregate-and-validate step for improving accuracy and determining a priority order. The refined rule-set is then loaded into an application-identification engine where it operates at a per flow granularity, in an extract-and-lookup paradigm, to identify the application responsible for a given flow. Thus, SAMPLES can facilitate important network measurement and management tasks --- e.g. behavioral profiling [29], application-level firewalls [21,22] etc. --- which require a more detailed view of the underlying traffic than that afforded by traditional protocol port based methods. We evaluate SAMPLES on a test set comprising 15 million flows (approx.) generated by over 700 K applications from the Android, iOS and Nokia market-places. SAMPLES successfully identifies over 90 of these applications with 99 accuracy on an average. This, in spite of the fact that fewer than 2 of the applications are required during the training phase, for each of the three market places. This is a testament to the universality and the scalability of our approach. We, therefore, expect SAMPLES to work with reasonable coverage and accuracy for other mobile platforms --- e.g. BlackBerry and Windows Mobile --- as well.",
"Smartphone users do not fully know what their apps do. For example, an applications' network usage and underlying security configuration is invisible to users. In this paper we introduce Securacy, a mobile app that explores users' privacy and security concerns with Android apps. Securacy takes a reactive, personalized approach, highlighting app permission settings that the user has previously stated are concerning, and provides feedback on the use of secure and insecure network communication for each app. We began our design of Securacy by conducting a literature review and in-depth interviews with 30 participants to understand their concerns. We used this knowledge to build Securacy and evaluated its use by another set of 218 anonymous participants who installed the application from the Google Play store. Our results show that access to address book information is by far the biggest privacy concern. Over half (56.4 ) of the connections made by apps are insecure, and the destination of the majority of network traffic is North America, regardless of the location of the user. Our app provides unprecedented insight into Android applications' communications behavior globally, indicating that the majority of apps currently use insecure network connections.",
"Abstract In this paper we present a new behavior-based anomaly detection system for detecting meaningful deviations in a mobile application's network behavior. The main goal of the proposed system is to protect mobile device users and cellular infrastructure companies from malicious applications by: (1) identification of malicious attacks or masquerading applications installed on a mobile device, and (2) identification of republished popular applications injected with a malicious code (i.e., repackaging). More specifically, we attempt to detect a new type of mobile malware with self-updating capabilities that were recently found on the official Google Android marketplace. Malware of this type cannot be detected using the standard signatures approach or by applying regular static or dynamic analysis methods. The detection is performed based on the application's network traffic patterns only. For each application, a model representing its specific traffic pattern is learned locally (i.e., on the device). Semi-supervised machine-learning methods are used for learning the normal behavioral patterns and for detecting deviations from the application's expected behavior. These methods were implemented and evaluated on Android devices. The evaluation experiments demonstrate that: (1) various applications have specific network traffic patterns and certain application categories can be distinguished by their network patterns; (2) different levels of deviation from normal behavior can be detected accurately; (3) in the case of self-updating malware, original (benign) and infected versions of an application have different and distinguishable network traffic patterns that in most cases, can be detected within a few minutes after the malware is executed while presenting very low false alarms rate; and (4) local learning is feasible and has a low performance overhead on mobile devices."
]
} |
1705.06629 | 2615361354 | Relying on ubiquitous Internet connectivity, applications on mobile devices frequently perform web requests during their execution. They fetch data for users to interact with, invoke remote functionalities, or send user-generated content or meta-data. These requests collectively reveal common practices of mobile application development, like what external services are used and how, and they point to possible negative effects like security and privacy violations, or impacts on battery life. In this paper, we assess different ways to analyze what web requests Android applications make. We start by presenting dynamic data collected from running 20 randomly selected Android applications and observing their network activity. Next, we present a static analysis tool, Stringoid, that analyzes string concatenations in Android applications to estimate constructed URL strings. Using Stringoid, we extract URLs from 30, 000 Android applications, and compare the performance with a simpler constant extraction analysis. Finally, we present a discussion of the advantages and limitations of dynamic and static analyses when extracting URLs, as we compare the data extracted by Stringoid from the same 20 applications with the dynamically collected data. | Beyond the here used Playdrone dataset of Android applications @cite_20 , other datasets have been published, that provide source code version histories of selected Android applications @cite_14 , or follow a continuous mining strategy to reflect the ever-growing Android ecosystem @cite_15 . We chose Playdrone when starting our work as the only dataset that made available the desired amount of application binaries. | {
"cite_N": [
"@cite_15",
"@cite_14",
"@cite_20"
],
"mid": [
"2407313496",
"2083342170",
""
],
"abstract": [
"We present a growing collection of Android Applications col-lected from several sources, including the official GooglePlay app market. Our dataset, AndroZoo, currently contains more than three million apps, each of which has beenanalysed by tens of different AntiVirus products to knowwhich applications are detected as Malware. We provide thisdataset to contribute to ongoing research efforts, as well asto enable new potential research topics on Android Apps.By releasing our dataset to the research community, we alsoaim at encouraging our fellow researchers to engage in reproducible experiments.",
"Android has grown to be the world's most popular mobile platform with apps that are capable of doing everything from checking sports scores to purchasing stocks. In order to assist researchers and developers in better understanding the development process as well as the current state of the apps themselves, we present a large dataset of analyzed open-source Android applications and provide a brief analysis of the data, demonstrating potential usefulness. This dataset contains 1,179 applications, including 4,416 different versions of these apps and 435,680 total commits. Furthermore, for each app we include the analytical results obtained from several static analysis tools including Androguard, Sonar, and Stowaway. In order to better support the community in conducting research on the security characteristics of the apps, our large analytical dataset comes with the detailed information including various versions of AndroidManifest.xml files and synthesized information such as permissions, intents, and minimum SDK. We collected 13,036 commits of the manifest files and recorded over 69,707 total permissions used. The results and a brief set of analytics are presented on our website: http: androsec.rit.edu.",
""
]
} |
1705.06631 | 2950897191 | The following game is played on a weighted graph: Alice selects a matching @math and Bob selects a number @math . Alice's payoff is the ratio of the weight of the @math heaviest edges of @math to the maximum weight of a matching of size at most @math . If @math guarantees a payoff of at least @math then it is called @math -robust. In 2002, Hassin and Rubinstein gave an algorithm that returns a @math -robust matching, which is best possible. We show that Alice can improve her payoff to @math by playing a randomized strategy. This result extends to a very general class of independence systems that includes matroid intersection, b-matchings, and strong 2-exchange systems. It also implies an improved approximation factor for a stochastic optimization variant known as the maximum priority matching problem and translates to an asymptotic robustness guarantee for deterministic matchings, in which Bob can only select numbers larger than a given constant. Moreover, we give a new LP-based proof of Hassin and Rubinstein's bound. | Results on the impact of randomization in robust optimization are much scarcer. Bertsimas, Nasrabadi, and Orlin @cite_5 show that the randomized version of network flow interdiction is equivalent to the maximum robust flow problem and use this insight to obtain an approximation algorithm for both problems. Mastin, Jaillet, and Chin @cite_10 consider a randomized version of the minmax (additive) regret model. They show that if an optimization problem can be solved efficiently, then also the corresponding randomized minmax regret version can be solved efficiently. Inspired by a preliminary version of our work, Kobayashi and Takazawa @cite_19 provide an analysis of the randomized robustness of general independence systems. They obtain a robustness guarantee of @math for general systems and a guarantee of @math for systems induced by instances of the knapsack problem (here @math is again the extendibility of the system, while @math is the ratio of the largest size of an independent set to the smallest size of a non-independent set minus @math ). They also construct instances that do not allow for a better than @math -robust or @math -robust randomized independent set, respectively. | {
"cite_N": [
"@cite_19",
"@cite_5",
"@cite_10"
],
"mid": [
"",
"1738845896",
"2153675120"
],
"abstract": [
"",
"In this paper, we introduce the randomized network interdiction problem that allows the interdictor to use randomness to select arcs to be removed. We model the problem in two different ways: arc-based and path-based formulations, depending on whether flows are defined on arcs or paths, respectively. We present insights into the modeling power, complexity, and approximability of both formulations.",
"The minmax regret problem for combinatorial optimization under uncertainty can be viewed as a zero-sum game played between an optimizing player and an adversary, where the optimizing player selects a solution and the adversary selects costs with the intention of maximizing the regret of the player. The conventional minmax regret model considers only deterministic solutions strategies, and minmax regret versions of most polynomial solvable problems are NP-hard. In this paper, we consider a randomized model where the optimizing player selects a probability distribution (corresponding to a mixed strategy) over solutions and the adversary selects costs with knowledge of the player’s distribution, but not its realization. We show that under this randomized model, the minmax regret version of any polynomial solvable combinatorial problem becomes polynomial solvable. This holds true for both interval and discrete scenario representations of uncertainty. Using the randomized model, we show new proofs of existing approximation algorithms for the deterministic model based on primal-dual approaches. We also determine integrality gaps of minmax regret formulations, giving tight bounds on the limits of performance gains from randomization. Finally, we prove that minmax regret problems are NP-hard under general convex uncertainty."
]
} |
1705.06821 | 2620126604 | The key idea of variational auto-encoders (VAEs) resembles that of traditional auto-encoder models in which spatial information is supposed to be explicitly encoded in the latent space. However, the latent variables in VAEs are vectors, which can be interpreted as multiple feature maps of size 1x1. Such representations can only convey spatial information implicitly when coupled with powerful decoders. In this work, we propose spatial VAEs that use feature maps of larger size as latent variables to explicitly capture spatial information. This is achieved by allowing the latent variables to be sampled from matrix-variate normal (MVN) distributions whose parameters are computed from the encoder network. To increase dependencies among locations on latent feature maps and reduce the number of parameters, we further propose spatial VAEs via low-rank MVN distributions. Experimental results show that the proposed spatial VAEs outperform original VAEs in capturing rich structural and spatial information. | Auto-encoder (AE) is a model architecture used in tasks like image segmentation @cite_20 @cite_27 @cite_21 , machine translation @cite_2 @cite_12 and denoising reconstruction @cite_11 @cite_29 . It consists of two parts: an encoder that encodes the input data into lower-dimensional latent representations and a decoder that generates outputs by decoding the representations. Depending on different tasks, the latent representations will focus on different properties of input data. Nevertheless, these tasks usually require outputs to have similar or exactly the same structure as inputs. Thus, structural information is expected to be preserved through the encoder-decoder process. | {
"cite_N": [
"@cite_29",
"@cite_21",
"@cite_27",
"@cite_2",
"@cite_20",
"@cite_12",
"@cite_11"
],
"mid": [
"2145094598",
"2952632681",
"1901129140",
"2133564696",
"",
"2949888546",
"2025768430"
],
"abstract": [
"We explore an original strategy for building deep networks, based on stacking layers of denoising autoencoders which are trained locally to denoise corrupted versions of their inputs. The resulting algorithm is a straightforward variation on the stacking of ordinary autoencoders. It is however shown on a benchmark of classification problems to yield significantly lower classification error, thus bridging the performance gap with deep belief networks (DBN), and in several cases surpassing it. Higher level representations learnt in this purely unsupervised fashion also help boost the performance of subsequent SVM classifiers. Qualitative experiments show that, contrary to ordinary autoencoders, denoising autoencoders are able to learn Gabor-like edge detectors from natural image patches and larger stroke detectors from digit images. This work clearly establishes the value of using a denoising criterion as a tractable unsupervised objective to guide the learning of useful higher level representations.",
"Convolutional networks are powerful visual models that yield hierarchies of features. We show that convolutional networks by themselves, trained end-to-end, pixels-to-pixels, exceed the state-of-the-art in semantic segmentation. Our key insight is to build \"fully convolutional\" networks that take input of arbitrary size and produce correspondingly-sized output with efficient inference and learning. We define and detail the space of fully convolutional networks, explain their application to spatially dense prediction tasks, and draw connections to prior models. We adapt contemporary classification networks (AlexNet, the VGG net, and GoogLeNet) into fully convolutional networks and transfer their learned representations by fine-tuning to the segmentation task. We then define a novel architecture that combines semantic information from a deep, coarse layer with appearance information from a shallow, fine layer to produce accurate and detailed segmentations. Our fully convolutional network achieves state-of-the-art segmentation of PASCAL VOC (20 relative improvement to 62.2 mean IU on 2012), NYUDv2, and SIFT Flow, while inference takes one third of a second for a typical image.",
"There is large consent that successful training of deep networks requires many thousand annotated training samples. In this paper, we present a network and training strategy that relies on the strong use of data augmentation to use the available annotated samples more efficiently. The architecture consists of a contracting path to capture context and a symmetric expanding path that enables precise localization. We show that such a network can be trained end-to-end from very few images and outperforms the prior best method (a sliding-window convolutional network) on the ISBI challenge for segmentation of neuronal structures in electron microscopic stacks. Using the same network trained on transmitted light microscopy images (phase contrast and DIC) we won the ISBI cell tracking challenge 2015 in these categories by a large margin. Moreover, the network is fast. Segmentation of a 512x512 image takes less than a second on a recent GPU. The full implementation (based on Caffe) and the trained networks are available at http: lmb.informatik.uni-freiburg.de people ronneber u-net .",
"Neural machine translation is a recently proposed approach to machine translation. Unlike the traditional statistical machine translation, the neural machine translation aims at building a single neural network that can be jointly tuned to maximize the translation performance. The models proposed recently for neural machine translation often belong to a family of encoder-decoders and consists of an encoder that encodes a source sentence into a fixed-length vector from which a decoder generates a translation. In this paper, we conjecture that the use of a fixed-length vector is a bottleneck in improving the performance of this basic encoder-decoder architecture, and propose to extend this by allowing a model to automatically (soft-)search for parts of a source sentence that are relevant to predicting a target word, without having to form these parts as a hard segment explicitly. With this new approach, we achieve a translation performance comparable to the existing state-of-the-art phrase-based system on the task of English-to-French translation. Furthermore, qualitative analysis reveals that the (soft-)alignments found by the model agree well with our intuition.",
"",
"Deep Neural Networks (DNNs) are powerful models that have achieved excellent performance on difficult learning tasks. Although DNNs work well whenever large labeled training sets are available, they cannot be used to map sequences to sequences. In this paper, we present a general end-to-end approach to sequence learning that makes minimal assumptions on the sequence structure. Our method uses a multilayered Long Short-Term Memory (LSTM) to map the input sequence to a vector of a fixed dimensionality, and then another deep LSTM to decode the target sequence from the vector. Our main result is that on an English to French translation task from the WMT'14 dataset, the translations produced by the LSTM achieve a BLEU score of 34.8 on the entire test set, where the LSTM's BLEU score was penalized on out-of-vocabulary words. Additionally, the LSTM did not have difficulty on long sentences. For comparison, a phrase-based SMT system achieves a BLEU score of 33.3 on the same dataset. When we used the LSTM to rerank the 1000 hypotheses produced by the aforementioned SMT system, its BLEU score increases to 36.5, which is close to the previous best result on this task. The LSTM also learned sensible phrase and sentence representations that are sensitive to word order and are relatively invariant to the active and the passive voice. Finally, we found that reversing the order of the words in all source sentences (but not target sentences) improved the LSTM's performance markedly, because doing so introduced many short term dependencies between the source and the target sentence which made the optimization problem easier.",
"Previous work has shown that the difficulties in learning deep generative or discriminative models can be overcome by an initial unsupervised learning step that maps inputs to useful intermediate representations. We introduce and motivate a new training principle for unsupervised learning of a representation based on the idea of making the learned representations robust to partial corruption of the input pattern. This approach can be used to train autoencoders, and these denoising autoencoders can be stacked to initialize deep architectures. The algorithm can be motivated from a manifold learning and information theoretic perspective or from a generative model perspective. Comparative experiments clearly show the surprising advantage of corrupting the input of autoencoders on a pattern classification benchmark suite."
]
} |
1705.06821 | 2620126604 | The key idea of variational auto-encoders (VAEs) resembles that of traditional auto-encoder models in which spatial information is supposed to be explicitly encoded in the latent space. However, the latent variables in VAEs are vectors, which can be interpreted as multiple feature maps of size 1x1. Such representations can only convey spatial information implicitly when coupled with powerful decoders. In this work, we propose spatial VAEs that use feature maps of larger size as latent variables to explicitly capture spatial information. This is achieved by allowing the latent variables to be sampled from matrix-variate normal (MVN) distributions whose parameters are computed from the encoder network. To increase dependencies among locations on latent feature maps and reduce the number of parameters, we further propose spatial VAEs via low-rank MVN distributions. Experimental results show that the proposed spatial VAEs outperform original VAEs in capturing rich structural and spatial information. | In computer vision tasks, structural information usually means spatial information of images. There are two main strategies to preserve spatial information in AE for image tasks. One is to apply very powerful decoders, like conditional pixel convolutional neural networks (PixelCNNs) @cite_5 @cite_23 @cite_25 @cite_10 , that generate output images pixel-by-pixel. In this way, the decoders can recover spatial information in the form of dependencies among pixels. However, pixel-by-pixel generation is very slow, resulting in major speed problems in practice. The other method is to let the latent representations explicitly contain spatial information and apply decoders that can make use of such information. To apply this strategy for image tasks, usually the latent representations are feature maps of size between the size of a pixel ( @math ) and that of the input image, while the decoders are deconvolutional neural networks (DCNNs) @cite_20 . Since most computer vision tasks only require high-level spatial information like relative locations of objects instead of detailed relationships among pixels, preserving only rough spatial information is enough, and this strategy is proved effective and efficient. | {
"cite_N": [
"@cite_23",
"@cite_5",
"@cite_10",
"@cite_25",
"@cite_20"
],
"mid": [
"2423557781",
"2953318193",
"2949899814",
"2581236139",
""
],
"abstract": [
"This work explores conditional image generation with a new image density model based on the PixelCNN architecture. The model can be conditioned on any vector, including descriptive labels or tags, or latent embeddings created by other networks. When conditioned on class labels from the ImageNet database, the model is able to generate diverse, realistic scenes representing distinct animals, objects, landscapes and structures. When conditioned on an embedding produced by a convolutional network given a single image of an unseen face, it generates a variety of new portraits of the same person with different facial expressions, poses and lighting conditions. We also show that conditional PixelCNN can serve as a powerful decoder in an image autoencoder. Additionally, the gated convolutional layers in the proposed model improve the log-likelihood of PixelCNN to match the state-of-the-art performance of PixelRNN on ImageNet, with greatly reduced computational cost.",
"Modeling the distribution of natural images is a landmark problem in unsupervised learning. This task requires an image model that is at once expressive, tractable and scalable. We present a deep neural network that sequentially predicts the pixels in an image along the two spatial dimensions. Our method models the discrete probability of the raw pixel values and encodes the complete set of dependencies in the image. Architectural novelties include fast two-dimensional recurrent layers and an effective use of residual connections in deep recurrent networks. We achieve log-likelihood scores on natural images that are considerably better than the previous state of the art. Our main results also provide benchmarks on the diverse ImageNet dataset. Samples generated from the model appear crisp, varied and globally coherent.",
"Natural image modeling is a landmark challenge of unsupervised learning. Variational Autoencoders (VAEs) learn a useful latent representation and model global structure well but have difficulty capturing small details. PixelCNN models details very well, but lacks a latent code and is difficult to scale for capturing large structures. We present PixelVAE, a VAE model with an autoregressive decoder based on PixelCNN. Our model requires very few expensive autoregressive layers compared to PixelCNN and learns latent codes that are more compressed than a standard VAE while still capturing most non-trivial structure. Finally, we extend our model to a hierarchy of latent variables at different scales. Our model achieves state-of-the-art performance on binarized MNIST, competitive performance on 64x64 ImageNet, and high-quality samples on the LSUN bedrooms dataset.",
"PixelCNNs are a recently proposed class of powerful generative models with tractable likelihood. Here we discuss our implementation of PixelCNNs which we make available at this https URL Our implementation contains a number of modifications to the original model that both simplify its structure and improve its performance. 1) We use a discretized logistic mixture likelihood on the pixels, rather than a 256-way softmax, which we find to speed up training. 2) We condition on whole pixels, rather than R G B sub-pixels, simplifying the model structure. 3) We use downsampling to efficiently capture structure at multiple resolutions. 4) We introduce additional short-cut connections to further speed up optimization. 5) We regularize the model using dropout. Finally, we present state-of-the-art log likelihood results on CIFAR-10 to demonstrate the usefulness of these modifications.",
""
]
} |
1705.06597 | 2951998065 | Polarization is a troubling phenomenon that can lead to societ al divisions and hurt the democratic process. It is therefore important to develop methods to reduce it. We propose an algorithmic solution to the problem of reducing polarization. The core idea is to expose users to content that challenges their point of view, with the hope broadening their perspective, and thus reduce their polarity. Our method takes into account several aspects of the problem, such as the estimated polarity of the user, the probability of accepting the recommendation, the polarity of the content, and popularity of the content being recommended. We evaluate our recommendations via a large-scale user study on Twitter users that were actively involved in the discussion of the US elections results. Results shows that, in most cases, the factors taken into account in the recommendation affect the users as expected, and thus capture the essential features of the problem. | using ideas from in order to make use of content from high degree (authoritative users) in the ranking and show the ranking. using ideas from we diversify the topics. using ideas from @cite_3 , we include an acceptance probability into our recommendation. | {
"cite_N": [
"@cite_3"
],
"mid": [
"1571597797"
],
"abstract": [
"ABSTRACT Social networks allow people to connect with each otherand have conversations on a wide variety of topics. How-ever, users tend to connect with like-minded people and readagreeable information, a behavior that leads to group polar-ization. Motivated by this scenario, we study how to takeadvantage of partial homophily to suggest agreeable contentto users authored by people with opposite views on sensitiveissues. We introduce a paradigm to present a data portraitof users, in which their characterizing topics are visualizedand their corresponding tweets are displayed using an or-ganic design. Among their tweets we inject recommendedtweets from other people considering their views on sensitiveissues in addition to topical relevance, indirectly motivatingconnections between dissimilar people. To evaluate our ap-proach, we present a case study on Twitter about a sensitivetopic in Chile, where we estimate user stances for regularpeople and nd intermediary topics. We then evaluated ourdesign in a user study. We found that recommending topi-cally relevant content from authors with opposite views in abaseline interface had a negative emotional e ect. We sawthat our organic visualization design reverts that e ect. Wealso observed signi cant individual di erences linked to eval-uation of recommendations. Our results suggest that organicvisualization may revert the negative e ects of providingpotentially sensitive content."
]
} |
1705.06840 | 2618915936 | Motivated by the common academic problem of allocating papers to referees for conference reviewing we propose a novel mechanism for solving the assignment problem when we have a two sided matching problem with preferences from one side (the agents reviewers) over the other side (the objects papers) and both sides have capacity constraints. The assignment problem is a fundamental problem in both computer science and economics with application in many areas including task and resource allocation. We draw inspiration from multi-criteria decision making and voting and use order weighted averages (OWAs) to propose a novel and flexible class of algorithms for the assignment problem. We show an algorithm for finding a @math -OWA assignment in polynomial time, in contrast to the NP-hardness of finding an egalitarian assignment. Inspired by this setting we observe an interesting connection between our model and the classic proportional multi-winner election problem in social choice. | The conference paper assignment has been studied a number of times over the years in computer science @cite_27 , as has defining and refining notions of fairness for the assignment vectors in multi-agent allocation problems @cite_25 . We build off the work of , who extensively study the notion of fair paper assignments, including lexi-min and rank-maximal assignments, within the context of conference paper assignment. show that for the setting we study, finding an egalitarian optimal assignment and finding a leximin optimal assignment are both NP-hard when there are three or more equivalence classes; and polynomial time computable when there are only two. They also provide an approximation algorithm for leximin optimal assignments. We know that if the capacity constraints are hard values, i.e., each reviewer must review @math papers and each paper must receive exactly @math reviews, then the resulting version of capacitated assignment is NP-hard @cite_3 . Answer set programming for CPAP was studied by ; they encode the CPAP problem in ASP and show that finding a solution that roughly correspond to the leximin optimal and egalitarian solutions can be done in reasonable time for large settings ( @math agents). | {
"cite_N": [
"@cite_27",
"@cite_25",
"@cite_3"
],
"mid": [
"2547627026",
"1599161738",
"2017082987"
],
"abstract": [
"Among the tasks to be carried out by conference organizers is the one of assigning reviewers to papers. That problem is known in the literature as the Conference Paper Assignment Problem CPAP. In this paper we approach the solution of a reasonably rich variant of the CPAP by means of Answer Set Programming ASP. ASP is an established logic-based programming paradigm which has been successfully applied for solving complex problems arising in Artificial Intelligence. We show how the CPAP can be elegantly encoded by means of an ASP program, and we analyze the results of an experiment, conducted on real-world data, that outlines the viability of our solution.",
"This paper deals with fair assignment problems in decision contexts involving multiple agents. In such problems, each agent has its own evaluation of costs and we want to find a fair compromise solution between individual point of views. Lorenz dominance is a standard decision model used in Economics to refine Pareto dominance while favoring solutions that fairly share happiness among agents. In order to enhance the discrimination possibilities offered by Lorenz dominance, we introduce here a new model called infinite order Lorenz dominance. We establish a representation result for this model using an ordered weighted average with decreasing weights. Hence we exhibit some properties of infinite order Lorenz dominance that explain how fairness is achieved in the aggregation of individual preferences. Then we explain how to solve fair assignment problems of m items to n agents, using infinite order Lorenz dominance and other models used for measuring inequalities. We show that this problem can be reformulated as a 0--1 non-linear optimization problems that can be solved, after a linearization step, by standard LP solvers. We provide numerical results showing the efficiency of the proposed approach on various instances of the paper assignment problem.",
"Peer review has become the most common practice for judging papers submitted to a conference for decades. An extremely important task involved in peer review is to assign submitted papers to reviewers with appropriate expertise which is referred to as paper-reviewer assignment. In this paper, we study the paper-reviewer assignment problem from both the goodness aspect and the fairness aspect. For the goodness aspect, we propose to maximize the topic coverage of the paper-reviewer assignment. This objective is new and the problem based on this objective is shown to be NP-hard. To solve this problem efficiently, we design an approximate algorithm which gives a 1 3-approximation. For the fairness aspect, we perform a detailed study on conflict-of-interest (COI) types and discuss several issues related to using COI, which, we hope, can raise some open discussions among researchers on the COI study. Finally, we conducted experiments on real datasets which verified the effectiveness of our algorithm and also revealed some interesting results of COI."
]
} |
1705.06840 | 2618915936 | Motivated by the common academic problem of allocating papers to referees for conference reviewing we propose a novel mechanism for solving the assignment problem when we have a two sided matching problem with preferences from one side (the agents reviewers) over the other side (the objects papers) and both sides have capacity constraints. The assignment problem is a fundamental problem in both computer science and economics with application in many areas including task and resource allocation. We draw inspiration from multi-criteria decision making and voting and use order weighted averages (OWAs) to propose a novel and flexible class of algorithms for the assignment problem. We show an algorithm for finding a @math -OWA assignment in polynomial time, in contrast to the NP-hardness of finding an egalitarian assignment. Inspired by this setting we observe an interesting connection between our model and the classic proportional multi-winner election problem in social choice. | CPAP also receives considerable attention in the recommender systems @cite_16 and machine learning @cite_7 communities. Often though, this work takes the approach of attempting to infer a more refined utility or preference model in order to distinguish papers. Fairness and efficiency concerns are secondary. A prime example of this is the Toronto Paper Matching System designed by . This system attempts to increase the accuracy of the matching algorithms by having the papers express preferences over the reviewers themselves; where these preferences are inferred from the contents of the papers. | {
"cite_N": [
"@cite_16",
"@cite_7"
],
"mid": [
"2126183471",
"1782140593"
],
"abstract": [
"We present a recommender systems approach to conference paper assignment, i.e., the task of assigning paper submissions to reviewers. We address both the modeling of reviewer-paper preferences (which can be cast as a learning problem) and the optimization of reviewing assignments to satisfy global conference criteria (which can be viewed as constraint satisfaction). Due to the paucity of preference data per reviewer or per paper (relative to other recommender systems applications) we show how we can integrate multiple sources of information to learn reviewer-paper preference models. Our models are evaluated not just in terms of prediction accuracy but in terms of end-assignment quality. Using a linear programming-based assignment optimization, we show how our approach better explores the space of unsupplied assignments to maximize the overall affinities of papers assigned to reviewers. We demonstrate our results on real reviewer bidding data from the IEEE ICDM 2007 conference.",
"At the heart of many scientific conferences is the problem of matching submitted papers to suitable reviewers. Arriving at a good assignment is a major and important challenge for any conference organizer. In this paper we propose a framework to optimize paper-to-reviewer assignments. Our framework uses suitability scores to measure pairwise affinity between papers and reviewers. We show how learning can be used to infer suitability scores from a small set of provided scores, thereby reducing the burden on reviewers and organizers. We frame the assignment problem as an integer program and propose several variations for the paper-to-reviewer matching domain. We also explore how learning and matching interact. Experiments on two conference data sets examine the performance of several learning methods as well as the effectiveness of the matching formulations."
]
} |
1705.06840 | 2618915936 | Motivated by the common academic problem of allocating papers to referees for conference reviewing we propose a novel mechanism for solving the assignment problem when we have a two sided matching problem with preferences from one side (the agents reviewers) over the other side (the objects papers) and both sides have capacity constraints. The assignment problem is a fundamental problem in both computer science and economics with application in many areas including task and resource allocation. We draw inspiration from multi-criteria decision making and voting and use order weighted averages (OWAs) to propose a novel and flexible class of algorithms for the assignment problem. We show an algorithm for finding a @math -OWA assignment in polynomial time, in contrast to the NP-hardness of finding an egalitarian assignment. Inspired by this setting we observe an interesting connection between our model and the classic proportional multi-winner election problem in social choice. | We make use of Order weighted averages (OWAs), often employed in multi-criteria decision making @cite_6 . OWAs have recently received attention in computational social choice for voting and ranking @cite_17 , finding a collective set of items for a group @cite_29 , and multi-winner voting with proportional representation @cite_14 @cite_8 . The key difference between CPAP and voting using OWAs in the ComSoc literature is that CPAP does not select a set of winners that all agents will share. Instead, all agents are allocated a possibly disjoint set of objects. | {
"cite_N": [
"@cite_14",
"@cite_8",
"@cite_29",
"@cite_6",
"@cite_17"
],
"mid": [
"2949245628",
"2293358884",
"2952061453",
"2060907774",
"2174198470"
],
"abstract": [
"The goal of this paper is to propose and study properties of multiwinner voting rules which can be consider as generalisations of single-winner scoring voting rules. We consider SNTV, Bloc, k-Borda, STV, and several variants of Chamberlin--Courant's and Monroe's rules and their approximations. We identify two broad natural classes of multiwinner score-based rules, and show that many of the existing rules can be captured by one or both of these approaches. We then formulate a number of desirable properties of multiwinner rules, and evaluate the rules we consider with respect to these properties.",
"Given a set of voters V, a set of candidates C, and voters' preferences over the candidates, multiwinner voting rules output a fixed-size subset of candidates committee. Under the Chamberlin---Courant multiwinner voting rule, one fixes a scoring vector of length |C|, and each voter's 'utility' for a given committee is defined to be the score that she assigns to her most preferred candidate in that committee; the goal is then to find a committee that maximizes the joint utility of all voters. The joint utility is typically identified either with the sum of all voters' utilities or with the utility of the least satisfied voter, resulting in, respectively, the utilitarian and the egalitarian variant of the Chamberlin---Courant's rule. For both of these cases, the problem of computing an optimal committee is NP-hard for general preferences, but becomes polynomial-time solvable if voters' preferences are single-peaked or single-crossing. In this paper, we propose a family of multiwinner voting rules that are based on the concept of ordered weighted average OWA and smoothly interpolate between the egalitarian and the utilitarian variants of the Chamberlin---Courant rule. We show that under moderate constraints on the weight vector we can recover many of the algorithmic results known for the egalitarian and the utilitarian version of Chamberlin---Courant's rule in this more general setting.",
"We consider the following problem: There is a set of items (e.g., movies) and a group of agents (e.g., passengers on a plane); each agent has some intrinsic utility for each of the items. Our goal is to pick a set of @math items that maximize the total derived utility of all the agents (i.e., in our example we are to pick @math movies that we put on the plane's entertainment system). However, the actual utility that an agent derives from a given item is only a fraction of its intrinsic one, and this fraction depends on how the agent ranks the item among the chosen, available, ones. We provide a formal specification of the model and provide concrete examples and settings where it is applicable. We show that the problem is hard in general, but we show a number of tractability results for its natural special cases.",
"The author is primarily concerned with the problem of aggregating multicriteria to form an overall decision function. He introduces a type of operator for aggregation called an ordered weighted aggregation (OWA) operator and investigates the properties of this operator. The OWA's performance is found to be between those obtained using the AND operator, which requires all criteria to be satisfied, and the OR operator, which requires at least one criteria to be satisfied. >",
"Positional scoring rules in voting compute the score of an alternative by summing the scores for the alternative induced by every vote. This summation principle ensures that all votes contribute equally to the score of an alternative. We relax this assumption and, instead, aggregate scores by taking into account the rank of a score in the ordered list of scores obtained from the votes. This defines a new family of voting rules, rank-dependent scoring rules (RDSRs), based on ordered weighted average (OWA) operators, which, include all scoring rules, and many others, most of which of new. We study some properties of these rules, and show, empirically, that certain RDSRs are less manipulable than Borda voting, across a variety of statistical cultures."
]
} |
1705.06130 | 2400517036 | In the context of the smart grid, we propose in this paper an algorithm that forms coalitions of agents, called prosumers, that both produce and consume. It is designed to be used by aggregators that aim at selling aggregated surplus of production of the prosumers they control. We rely on real weather data sampled across stations of a given territory in order to simulate realistic production and consumption patterns for each prosumer. This enables us to capture geographical correlations among the agents while preserving the diversity due to different behaviors. As aggregators are bound to the market operator by a contract, they seek to maximize their offer while minimizing their risk. The proposed graph-based algorithm takes the underlying correlation structure of the agents into account and outputs coalitions with both high productivity and low variability. We show that the resulting diversified coalitions are able to generate higher benefits on a constrained energy market, and are more resilient to random failures of the agents. | Dynamic pricing is an interesting and useful tool, but it has also some limitations. It is likely that dynamic pricing will serve as a shaping mass tool while finer techniques will be needed locally. A popular approach in this direction consists in deploying storage devices in the network and using them as electricity buffers. Basically these storages would be charged when there is a surplus of production, and discharged when the consumption exceeds the production. Although quite simple, this idea causes numerous challenges when it comes to its real implementation. As long as the system considered remains small, a centralized control of these buffers could be envisioned. However, for real large systems it is likely that more sophisticated decentralized algorithms will be necessary. In @cite_8 , the authors introduce a distributed energy management system with a high use of renewables such that power is scheduled in a distributed fashion. In @cite_17 the optimal storage capacity problem is addressed. There is indeed an interesting trade-off between the costs of the equipments and the expected availability of power. The authors develop a framework that enables them to exhibit a Pareto front of efficient solutions. | {
"cite_N": [
"@cite_17",
"@cite_8"
],
"mid": [
"2068422870",
"2080637564"
],
"abstract": [
"Renewable energy sources continues to gain popularity. However, two major limitations exist that prevent widespread adoption: availability of the electricity generated and the cost of the equipment. Distributed generation, (DG) grid-tied photovoltaic-wind hybrid systems with centralized battery back-up, can help mitigate the variability of the renewable energy resource. The downside, however, is the cost of the equipment needed to create such a system. Thus, optimization of generation and storage in light of capital cost and variability mitigation is imperative to the financial feasibility of DC microgrid systems. PV and wind generation are both time dependent and variable but are highly correlated, which make them ideal for a dual-sourced hybrid system. This paper presents an optimization technique base on a Multi-Objective Genetic Algorithm (MOGA) which uses high temporal resolution insolation data taken at 10 seconds data rate instead of more commonly used hourly data rate. The proposed methodology employs a techno-economic approach to determine the system design optimized by considering multiple criteria including size, cost, and availability. The result is the baseline system cost necessary to meet the load requirements and which can also be used to monetize ancillary services that the smart DC microgrid can provide to the utility at the point of common coupling (PCC) such as voltage regulation. The hybrid smart DC microgrid community system optimized using high-temporal resolution data is compared to a system optimized using lower-rate temporal data to examine the effect of the temporal sampling of the renewable energy resource.",
"Due to its reduced communication overhead and robustness to failures, distributed energy management is of paramount importance in smart grids, especially in microgrids, which feature distributed generation (DG) and distributed storage (DS). Distributed economic dispatch for a microgrid with high renewable energy penetration and demand-side management operating in grid-connected mode is considered in this paper. To address the intrinsically stochastic availability of renewable energy sources (RES), a novel power scheduling approach is introduced. The approach involves the actual renewable energy as well as the energy traded with the main grid, so that the supply-demand balance is maintained. The optimal scheduling strategy minimizes the microgrid net cost, which includes DG and DS costs, utility of dispatchable loads, and worst-case transaction cost stemming from the uncertainty in RES. Leveraging the dual decomposition, the optimization problem formulated is solved in a distributed fashion by the local controllers of DG, DS, and dispatchable loads. Numerical results are reported to corroborate the effectiveness of the novel approach."
]
} |
1705.06130 | 2400517036 | In the context of the smart grid, we propose in this paper an algorithm that forms coalitions of agents, called prosumers, that both produce and consume. It is designed to be used by aggregators that aim at selling aggregated surplus of production of the prosumers they control. We rely on real weather data sampled across stations of a given territory in order to simulate realistic production and consumption patterns for each prosumer. This enables us to capture geographical correlations among the agents while preserving the diversity due to different behaviors. As aggregators are bound to the market operator by a contract, they seek to maximize their offer while minimizing their risk. The proposed graph-based algorithm takes the underlying correlation structure of the agents into account and outputs coalitions with both high productivity and low variability. We show that the resulting diversified coalitions are able to generate higher benefits on a constrained energy market, and are more resilient to random failures of the agents. | Nevertheless, one of the key point in the Markowitz theory is to consider explicitly the correlation between the assets since they impact directly the variances of the portfolios. Since the work of @cite_5 , an interesting approach consists in computing a distance metric based on the correlation coefficients in order to organize the series in a correlation graph. Nodes represent the series considered while the edges are weighted by the metric. Because the metric can be computed for all pairs, these graphs are complete and of little use as is. Historically, the approach used by @cite_5 was to compute a minimum spanning tree as to obtain a hierarchical clustering of the series. Later on, it was pointed out that, by definition, a spanning tree could not capture the underlying clustering structure hidden in the correlation graph. In this paper, we use another classical filtering technique called @math -graph @cite_1 . It consists in selecting a threshold @math , and filtering out edges with smaller weights. As we will see further in this paper, this approach has the advantage of preserving clusters of correlated series. | {
"cite_N": [
"@cite_5",
"@cite_1"
],
"mid": [
"2138144286",
"2131841184"
],
"abstract": [
"I find a hierarchical arrangement of stocks traded in a financial market by investigating the daily time series of the logarithm of stock price. The topological space is a subdominant ultrametric space associated with a graph connecting the stocks of the portfolio analyzed. The graph is obtained starting from the matrix of correlation coefficient computed between all pairs of stocks of the portfolio by considering the synchronous time evolution of the difference of the logarithm of daily stock price. The hierarchical tree of the subdominant ultrametric space associated with the graph provides a meaningful economic taxonomy.",
"We investigate the properties of correlation based networks originating from economic complex systems, such as the network of stocks traded at the New York Stock Exchange (NYSE). The weaker links (low correlation) of the system are found to contribute to the overall connectivity of the network significantly more than the strong links (high correlation). We find that nodes connected through strong links form well defined communities. These communities are clustered together in more complex ways compared to the widely used classification according to the economic activity. We find that some companies, such as General Electric (GE), Coca Cola (KO), and others, can be involved in different communities. The communities are found to be quite stable over time. Similar results were obtained by investigating markets completely different in size and properties, such as the Athens Stock Exchange (ASE). The present method may be also useful for other networks generated through correlations."
]
} |
1705.06173 | 2969859287 | We give fault-tolerant algorithms for establishing synchrony in distributed systems in which each of the n nodes has its own clock. Our algorithms operate in a very strong fault model: we require self-stabilisation, i.e., the initial state of the system may be arbitrary, and there can be up to f< n 3 ongoing Byzantine faults, i.e., nodes that deviate from the protocol in an arbitrary manner. Furthermore, we assume that the local clocks of the nodes may progress at different speeds (clock drift) and communication has bounded delay. In this model, we study the pulse synchronisation problem, where the task is to guarantee that eventually all correct nodes generate well-separated local pulse events (i.e., unlabelled logical clock ticks) in a synchronised manner. Compared to prior work, we achieve exponential improvements in stabilisation time and the number of communicated bits, and give the first sublinear-time algorithm for the problem: • In the deterministic setting, the state-of-the-art solutions stabilise in time Θ (f) and have each node broadcast Θ(f log f) bits per time unit. We exponentially reduce the number of bits broadcasted per time unit to Θ (log f) while retaining the same stabilisation time. • In the randomised setting, the state-of-the-art solutions stabilise in time Θ(f) and have each node broadcast O(1) bits per time unit. We exponentially reduce the stabilisation time to polylog f while each node broadcasts polylog f bits per time unit. These results are obtained by means of a recursive approach reducing the above task of self-stabilising pulse synchronisation in the bounded-delay model to non-self-stabilising binary consensus in the synchronous model. In general, our approach introduces at most logarithmic overheads in terms of stabilisation time and broadcasted bits over the underlying consensus routine. | If one takes any one of the elements described above out of the picture, then this greatly simplifies the problem. Without permanent ongoing faults, the problem becomes trivial: it suffices to have all nodes follow a designated leader. Without transient faults @cite_19 , straightforward solutions are given by elegant classics @cite_29 @cite_13 , where @cite_13 also guarantees asymptotically optimal skew @cite_16 . Taking the uncertainty of unknown message delays and drifting clocks out of the equation leads to the so-called digital clock synchronisation problem @cite_15 @cite_28 @cite_11 @cite_6 , where communication proceeds in synchronous rounds and the task is to agree on a consistent (bounded) round counter. While this abstraction is unrealistic as a basic system model, it yields conceptual insights into the pulse synchronisation problem in the bounded-delay model. Moreover, it is useful to assign numbers to pulses after pulse synchronisation is solved, in order to get a fully-fledged shared system-wide clock @cite_23 . | {
"cite_N": [
"@cite_28",
"@cite_29",
"@cite_6",
"@cite_19",
"@cite_23",
"@cite_15",
"@cite_16",
"@cite_13",
"@cite_11"
],
"mid": [
"1911929785",
"",
"2949609009",
"2118891797",
"2051004389",
"1997521442",
"2035918100",
"1994441984",
"2623176784"
],
"abstract": [
"Consider a complete communication network on n nodes. In synchronous 2-counting, the nodes receive a common clock pulse and they have to agree on which pulses are \"odd\" and which are \"even\". Furthermore, the solution needs to be self-stabilising (reaching correct operation from any initial state) and tolerate f Byzantine failures (nodes that send arbitrary misinformation). Prior algorithms either require a source of random bits or a large number of states per node. In this work, we give fast state-optimal deterministic algorithms for the first non-trivial case f = 1 . To obtain these algorithms, we develop and evaluate two different techniques for algorithm synthesis. Both are based on casting the synthesis problem as a propositional satisfiability (SAT) problem; a direct encoding is efficient for synthesising time-optimal algorithms, while an approach based on counter-example guided abstraction refinement discovers non-optimal algorithms quickly. We develop computational techniques to find algorithms for synchronous 2-counting.Automated synthesis yields state-optimal self-stabilising fault-tolerant algorithms.We give a thorough experimental comparison of our two SAT-based synthesis techniques.A direct SAT encoding is more efficient for finding time-optimal algorithms.An iterative CEGAR-based approach finds non-optimal algorithms quickly.",
"",
"Consider a fully-connected synchronous distributed system consisting of @math nodes, where up to @math nodes may be faulty and every node starts in an arbitrary initial state. In the synchronous @math -counting problem, all nodes need to eventually agree on a counter that is increased by one modulo @math in each round for given @math . In the self-stabilising firing squad problem, the task is to eventually guarantee that all non-faulty nodes have simultaneous responses to external inputs: if a subset of the correct nodes receive an external \"go\" signal as input, then all correct nodes should agree on a round (in the not-too-distant future) in which to jointly output a \"fire\" signal. Moreover, no node should generate a \"fire\" signal without some correct node having previously received a \"go\" signal as input. We present a framework reducing both tasks to binary consensus at very small cost. For example, we obtain a deterministic algorithm for self-stabilising Byzantine firing squads with optimal resilience @math , asymptotically optimal stabilisation and response time @math , and message size @math . As our framework does not restrict the type of consensus routines used, we also obtain efficient randomised solutions, and it is straightforward to adapt our framework for other types of permanent faults.",
"Algorithms are described for maintaining clock synchrony in a distributed multiprocess system where each process has its own clock. These algorithms work in the presence of arbitrary clock or process failures, including “two-faced clocks” that present different values to different processes. Two of the algorithms require that fewer than one-third of the processes be faulty. A third algorithm works if fewer than half the processes are faulty, but requires digital signatures.",
"In this paper, we show how to build synchronized clocks of arbitrary size atop of existing small-sized clocks, despite arbitrary faults. Our solution is both self-stabilizing and Byzantine fault-tolerant, and needs merely single-bit channels. It involves a reduction to Byzantine fault-tolerant consensus, which allows different consensus algorithms to be plugged in for matching the actual clock sizes and resilience requirements best. We demonstrate the practicability of our approach by means of an FPGA implementation and its experimental evaluation. To also address the cases where deterministic algorithms hit fundamental limits, we provide a novel randomized self-stabilizing Byzantine consensus algorithm that works very well also in these settings, along with its correctness proof and stabilization time analysis.",
"Consider a distributed network in which up to a third of the nodes may be Byzantine, and in which the non-faulty nodes may be subject to transient faults that alter their memory in an arbitrary fashion. Within the context of this model, we are interested in the digital clock synchronization problem; which consists of agreeing on bounded integer counters, and increasing these counters regularly. It has been postulated in the past that synchronization cannot be solved in a Byzantine tolerant and self-stabilizing manner. The first solution to this problem had an expected exponential convergence time. Later, a deterministic solution was published with linear convergence time, which is optimal for deterministic solutions. In the current paper we achieve an expected constant convergence time. We thus obtain the optimal probabilistic solution, both in terms of convergence time and in terms of resilience to Byzantine adversaries.",
"The problem of synchronizing clocks of processes in a fully connected network is considered. It is proved that, even if the clocks all run at the same rate as real time and there are no failures, an uncertainty of e in the message delivery time makes it impossible to synchronize the clocks of n processes any more closely than e(1−1 n ). A simple algorithm is given that achieves this bound.",
"Abstract We describe a new fault-tolerant algorithm for solving a variant of Lamport's clock synchronization problem. The algorithm is designed for a system of distributed processes that communicate by sending messages. Each process has its own read-only physical clock whose drift rate from real time is very small. By adding a value to its physical clock time, the process obtaines its local time. The algorithm solves the problem of maintaining closely synchronized local times, assuming that processes' local times are closely synchronized initially. The algorithm is able to tolerate the failure of just under one-third of the participating processes. It maintains synchronization to within a small constant, whose magnitude depends upon the rate of clock drift, the message delivery time and its uncertainty, and the initial closeness of synchronization. We also give a characterization of how far the clocks drift from real time. Reintegration of a repaired process can be accomplished using a slight modification of the basic alborithm. A similar style algorithm can also be used to achieve synchronization initially.",
"Consider a complete communication network of @math nodes, where the nodes receive a common clock pulse. We study the synchronous @math -counting problem: given any starting state and up to @math faulty nodes with arbitrary behavior, the task is to eventually have all correct nodes labeling the pulses with increasing values modulo @math in agreement. Thus, we are considering algorithms that are self-stabilizing despite Byzantine failures. In this work, we give new algorithms for the synchronous counting problem that (1) are deterministic, (2) have optimal resilience, (3) have a linear stabilization time in @math (asymptotically optimal), (4) use a small number of states, and, consequently, (5) communicate a small number of bits per round. Prior algorithms either resort to randomization, use a large number of states and need high communication bandwidth, or have suboptimal resilience. In particular, we achieve an exponential improvement in both state complexity and message size for deterministic algorithms. Moreover, ..."
]
} |
1705.06173 | 2969859287 | We give fault-tolerant algorithms for establishing synchrony in distributed systems in which each of the n nodes has its own clock. Our algorithms operate in a very strong fault model: we require self-stabilisation, i.e., the initial state of the system may be arbitrary, and there can be up to f< n 3 ongoing Byzantine faults, i.e., nodes that deviate from the protocol in an arbitrary manner. Furthermore, we assume that the local clocks of the nodes may progress at different speeds (clock drift) and communication has bounded delay. In this model, we study the pulse synchronisation problem, where the task is to guarantee that eventually all correct nodes generate well-separated local pulse events (i.e., unlabelled logical clock ticks) in a synchronised manner. Compared to prior work, we achieve exponential improvements in stabilisation time and the number of communicated bits, and give the first sublinear-time algorithm for the problem: • In the deterministic setting, the state-of-the-art solutions stabilise in time Θ (f) and have each node broadcast Θ(f log f) bits per time unit. We exponentially reduce the number of bits broadcasted per time unit to Θ (log f) while retaining the same stabilisation time. • In the randomised setting, the state-of-the-art solutions stabilise in time Θ(f) and have each node broadcast O(1) bits per time unit. We exponentially reduce the stabilisation time to polylog f while each node broadcasts polylog f bits per time unit. These results are obtained by means of a recursive approach reducing the above task of self-stabilising pulse synchronisation in the bounded-delay model to non-self-stabilising binary consensus in the synchronous model. In general, our approach introduces at most logarithmic overheads in terms of stabilisation time and broadcasted bits over the underlying consensus routine. | In contrast to these relaxed problem formulations, the pulse synchronisation problem was initially considered to be very challenging -- if not impossible -- to solve. In a seminal article, Dolev and Welch @cite_17 proved otherwise, albeit with an algorithm having an impractical exponential stabilisation time. In a subsequent line of work, the stabilisation time was reduced to polynomial @cite_14 and then linear in @math @cite_2 . However, the linear-time algorithm relies on simulating multiple instances of synchronous algorithms @cite_4 concurrently, which results in a high communication complexity. | {
"cite_N": [
"@cite_14",
"@cite_2",
"@cite_4",
"@cite_17"
],
"mid": [
"1849996689",
"1548544878",
"2126924915",
"1970745521"
],
"abstract": [
"We define the \"Pulse Synchronization\" problem that requires nodes to achieve tight synchronization of regular pulse events, in the settings of distributed computing systems. Pulse-coupled synchronization is a phenomenon displayed by a large variety of biological systems, typically overcoming a high level of noise. Inspired by such biological models, a robust and self-stabilizing pulse synchronization algorithm for distributed computer systems is presented. The algorithm attains near optimal synchronization tightness while tolerating up to a third of the nodes exhibiting Byzantine behavior concurrently. We propose that pulse synchronization algorithms can be suitable for a variety of distributed tasks that require tight synchronization but which can tolerate a bound variation in the regularity of the synchronized pulse invocations.",
"\"Pulse Synchronization\" intends to invoke a recurring distributed event at the different nodes, of a distributed system as simultaneously as possible and with a frequency that matches a predetermined regularity. This paper shows how to achieve that goal when the system is facing both transient and permanent (Byzantine) failures. Byzantine nodes might incessantly try to de-synchronize the correct nodes. Transient failures might throw the system into an arbitrary state in which correct nodes have no common notion what-so-ever, such as time or round numbers, and thus cannot use any aspect of their own local states to infer anything about the states of other correct nodes. The algorithm we present here guarantees that eventually all correct nodes will invoke their pulses within a very short time interval of each other and will do so regularly. The problem of pulse synchronization was recently solved in a system in which there exists an outside beat system that synchronously signals all nodes at once. In this paper we present a solution for a bounded-delay system. When the system in a steady state, a message sent by a correct node arrives and is processed by all correct nodes within a bounded time, say d time units, where at steady state the number of Byzantine nodes, f, should obey the n > 3f inequality, for a network of n nodes.",
"The problem addressed here concerns a set of isolated processors, some unknown subset of which may be faulty, that communicate only by means of two-party messages. Each nonfaulty processor has a private value of information that must be communicated to each other nonfaulty processor. Nonfaulty processors always communicate honestly, whereas faulty processors may lie. The problem is to devise an algorithm in which processors communicate their own values and relay values received from others that allows each nonfaulty processor to infer a value for each other processor. The value inferred for a nonfaulty processor must be that processor's private value, and the value inferred for a faulty one must be consistent with the corresponding value inferred by each other nonfaulty processor. It is shown that the problem is solvable for, and only for, n ≥ 3 m + 1, where m is the number of faulty processors and n is the total number. It is also shown that if faulty processors can refuse to pass on information but cannot falsely relay information, the problem is solvable for arbitrary n ≥ m ≥ 0. This weaker assumption can be approximated in practice using cryptographic methods.",
"We initiate a study of bounded clock synchronization under a more severe fault model than that proposed by Lamport and Melliar-Smith [1985]. Realistic aspects of the problem of synchronizing clocks in the presence of faults are considered. One aspect is that clock synchronization is an on-going task, thus the assumption that some of the processors never fail is too optimistic. To cope with this reality, we suggest self-stabilizing protocols that stabilize in any (long enough) period in which less than a third of the processors are faulty. Another aspect is that the clock value of each processor is bounded. A single transient fault may cause the clock to reach the upper bound. Therefore, we suggest a bounded clock that wraps around when appropriate.We present two randomized self-stabilizing protocols for synchronizing bounded clocks in the presence of Byzantine processor failures. The first protocol assumes that processors have a common pulse, while the second protocol does not. A new type of distributed counter based on the Chinese remainder theorem is used as part of the first protocol."
]
} |
1705.06173 | 2969859287 | We give fault-tolerant algorithms for establishing synchrony in distributed systems in which each of the n nodes has its own clock. Our algorithms operate in a very strong fault model: we require self-stabilisation, i.e., the initial state of the system may be arbitrary, and there can be up to f< n 3 ongoing Byzantine faults, i.e., nodes that deviate from the protocol in an arbitrary manner. Furthermore, we assume that the local clocks of the nodes may progress at different speeds (clock drift) and communication has bounded delay. In this model, we study the pulse synchronisation problem, where the task is to guarantee that eventually all correct nodes generate well-separated local pulse events (i.e., unlabelled logical clock ticks) in a synchronised manner. Compared to prior work, we achieve exponential improvements in stabilisation time and the number of communicated bits, and give the first sublinear-time algorithm for the problem: • In the deterministic setting, the state-of-the-art solutions stabilise in time Θ (f) and have each node broadcast Θ(f log f) bits per time unit. We exponentially reduce the number of bits broadcasted per time unit to Θ (log f) while retaining the same stabilisation time. • In the randomised setting, the state-of-the-art solutions stabilise in time Θ(f) and have each node broadcast O(1) bits per time unit. We exponentially reduce the stabilisation time to polylog f while each node broadcasts polylog f bits per time unit. These results are obtained by means of a recursive approach reducing the above task of self-stabilising pulse synchronisation in the bounded-delay model to non-self-stabilising binary consensus in the synchronous model. In general, our approach introduces at most logarithmic overheads in terms of stabilisation time and broadcasted bits over the underlying consensus routine. | The linear-time pulse synchronisation algorithm in @cite_2 relies on simulating (up to) one synchronous consensus instance for each node simultaneously. Accordingly, this protocol requires each node to broadcast @math bits per time unit. Moreover, the use of consensus is crucial, as failure of any consensus instance to generate correct output within a prespecified time bound may result in loss of synchrony, i.e., the algorithm would fail apparent stabilisation. @cite_5 , these obstacles were overcome by avoiding the use of consensus by reducing the pulse synchronisation problem to the easier task of generating at least one well-separated resynchronisation point'', which is roughly uniformly distributed within any period of @math time. This can be achieved by trying to initiate such a resynchronisation point at random times, in combination with threshold voting and locally checked timing constraints to rein in the influence of Byzantine nodes. In a way, this seems much simpler than solving consensus, but the randomisation used to obtain a suitable resynchronisation point strongly reminds of the power provided by shared coins @cite_8 @cite_26 @cite_9 @cite_15 -- and this is exactly what the core routine of the expected constant-round consensus algorithm from @cite_9 provides. | {
"cite_N": [
"@cite_26",
"@cite_8",
"@cite_9",
"@cite_2",
"@cite_5",
"@cite_15"
],
"mid": [
"1549384320",
"1991153938",
"1483193990",
"1548544878",
"2110142332",
"1997521442"
],
"abstract": [
"",
"We present a randomized solution for the Byzantine Generals Problems. The solution works in the synchronous as well as the asynchronous case and produces Byzantine Agreement within a fixed small expected number of computational rounds, independent of the number n of processes and the bound t on the number of faulty processes. The solution uses A. Shamir's method for sharing secrets. It specializes to produce a simple solution for the Distributed Commit problem.",
"",
"\"Pulse Synchronization\" intends to invoke a recurring distributed event at the different nodes, of a distributed system as simultaneously as possible and with a frequency that matches a predetermined regularity. This paper shows how to achieve that goal when the system is facing both transient and permanent (Byzantine) failures. Byzantine nodes might incessantly try to de-synchronize the correct nodes. Transient failures might throw the system into an arbitrary state in which correct nodes have no common notion what-so-ever, such as time or round numbers, and thus cannot use any aspect of their own local states to infer anything about the states of other correct nodes. The algorithm we present here guarantees that eventually all correct nodes will invoke their pulses within a very short time interval of each other and will do so regularly. The problem of pulse synchronization was recently solved in a system in which there exists an outside beat system that synchronously signals all nodes at once. In this paper we present a solution for a bounded-delay system. When the system in a steady state, a message sent by a correct node arrives and is processed by all correct nodes within a bounded time, say d time units, where at steady state the number of Byzantine nodes, f, should obey the n > 3f inequality, for a network of n nodes.",
"Today’s hardware technology presents a new challenge in designing robust systems. Deep submicron VLSI technology introduces transient and permanent faults that were never considered in low-level system designs in the past. Still, robustness of that part of the system is crucial and needs to be guaranteed for any successful product. Distributed systems, on the other hand, have been dealing with similar issues for decades. However, neither the basic abstractions nor the complexity of contemporary fault-tolerant distributed algorithms match the peculiarities of hardware implementations. This article is intended to be part of an attempt striving to bridge over this gap between theory and practice for the clock synchronization problem. Solving this task sufficiently well will allow to build an ultra-robust high-precision clocking system for hardware designs like systems-on-chips in critical applications. As our first building block, we describe and prove correct a novel distributed, Byzantine fault-tolerant, probabilistically self-stabilizing pulse synchronization protocol, called FATAL, that can be implemented using standard asynchronous digital logic: Correct FATAL nodes are guaranteed to generate pulses (i.e., unnumbered clock ticks) in a synchronized way, despite a certain fraction of nodes being faulty. FATAL uses randomization only during stabilization and, despite the strict limitations introduced by hardware designs, offers optimal resilience and smaller complexity than all existing protocols. Finally, we show how to leverage FATAL to efficiently generate synchronized, self-stabilizing, high-frequency clocks.",
"Consider a distributed network in which up to a third of the nodes may be Byzantine, and in which the non-faulty nodes may be subject to transient faults that alter their memory in an arbitrary fashion. Within the context of this model, we are interested in the digital clock synchronization problem; which consists of agreeing on bounded integer counters, and increasing these counters regularly. It has been postulated in the past that synchronization cannot be solved in a Byzantine tolerant and self-stabilizing manner. The first solution to this problem had an expected exponential convergence time. Later, a deterministic solution was published with linear convergence time, which is optimal for deterministic solutions. In the current paper we achieve an expected constant convergence time. We thus obtain the optimal probabilistic solution, both in terms of convergence time and in terms of resilience to Byzantine adversaries."
]
} |
1705.06135 | 2616586468 | Answering queries over a federation of SPARQL endpoints requires combining data from more than one data source. Optimizing queries in such scenarios is particularly challenging not only because of (i) the large variety of possible query execution plans that correctly answer the query but also because (ii) there is only limited access to statistics about schema and instance data of remote sources. To overcome these challenges, most federated query engines rely on heuristics to reduce the space of possible query execution plans or on dynamic programming strategies to produce optimal plans. Nevertheless, these plans may still exhibit a high number of intermediate results or high execution times because of heuristics and inaccurate cost estimations. In this paper, we present Odyssey, an approach that uses statistics that allow for a more accurate cost estimation for federated queries and therefore enables Odyssey to produce better query execution plans. Our experimental results show that Odyssey produces query execution plans that are better in terms of data transfer and execution time than state-of-the-art optimizers. Our experiments using the FedBench benchmark show execution time gains of at least 25 times on average. | Federated query optimization can also rely on cardinality estimations based on statistics and used, for instance, to reduce sizes of intermediate results. Most available statistics @cite_16 use the Vocabulary of Interlinked Datasets voiD @cite_6 , which describes statistics at dataset level (e.g., the number of triples), at the property level (e.g., for each property, its number of different subjects), and at the class level (e.g., the number of instances of each class). However, approaches based on voiD @cite_8 @cite_10 @cite_0 and other statistics, such as QTrees @cite_20 and PARTrees @cite_5 , share the drawback of missing the best query execution plans because of errors in estimating cardinalities caused by relying on assumptions that often do not hold for arbitrary RDF datasets @cite_19 , e.g., a uniform data distribution and that the results of triple patterns are independent. | {
"cite_N": [
"@cite_8",
"@cite_6",
"@cite_0",
"@cite_19",
"@cite_5",
"@cite_16",
"@cite_10",
"@cite_20"
],
"mid": [
"2007786749",
"156172657",
"2265585838",
"2100387739",
"2051493734",
"1898874593",
"1842364016",
"2034103898"
],
"abstract": [
"Processing SPARQL queries involves the construction of an efficient query plan to guide query execution. Alternative plans can vary in the resources and the amount of time that they need by orders of magnitude, making planning crucial for efficiency. On the other hand, the construction of optimal plans can become computationally intensive and it also operates upon detailed, difficult to obtain, metadata. In this paper we present Semagrow, a federated SPARQL querying system that uses metadata about the federated data sources in order to optimize query execution. We balance between a query optimizer that introduces little overhead, has appropriate fall backs in the absence of metadata, but at the same time produces optimal plans in as many situations as possible. Semagrow also exploits non-blocking and asynchronous stream processing technologies to achieve query execution efficiency and robustness. We also present and analyse empirical results using the FedBench benchmark to compare Semagrow against FedX and SPLENDID. Semagrow clearly outperforms SPLENDID and it is either on a par or much faster than FedX.",
"",
"In order to leverage the full potential of the Semantic Web it is necessary to transparently query distributed RDF data sources in the same way as it has been possible with federated databases for ages. However, there are significant differences between the Web of (linked) Data and the traditional database approaches. Hence, it is not straightforward to adapt successful database techniques for RDF federation. Reasons are the missing cooperation between SPARQL end-points and the need for detailed data statistics for estimating the costs of query execution plans. We have implemented SPLENDID, a query optimization strategy for federating SPARQL endpoints based on statistical data obtained from voiD descriptions.",
"Accurate cardinality estimates are essential for a successful query optimization. This is not only true for relational DBMSs but also for RDF stores. An RDF database consists of a set of triples and, hence, can be seen as a relational database with a single table with three attributes. This makes RDF rather special in that queries typically contain many self joins. We show that relational DBMSs are not well-prepared to perform cardinality estimation in this context. Further, there are hardly any special cardinality estimation methods for RDF databases. To overcome this lack of appropriate cardinality estimation methods, we introduce characteristic sets together with new cardinality estimation methods based upon them. We then show experimentally that the new methods are-in the RDF context-highly superior to the estimation methods employed by commercial DBMSs and by the open-source RDF store RDF-3X.",
"The inherent flexibility of the RDF data model has led to its notable adoption in many domains, especially in the area of life-sciences. Some of these domains have an emerging need to access data integrated from various distributed sources of information. It is not always possible to implement this by simply loading all data into one central RDF store. For example, in the context of inter-institutional collaboration for drug development and clinical research participants often want to maintain control over their local databases. Alternatively, distributed query processing techniques can be utilized to evaluate queries by accessing the remote data sources only on demand and in conformance with local authorization models. In this paper we present an efficient approach to distributed query processing for large autonomous RDF databases. The groundwork is laid by a comprehensive RDF-specific schema- and instance-level synopsis. We present an optimizer that is able to utilize this synopsis to generate compact execution plans by precisely determining, at compile-time, those sources that are relevant to a query. Furthermore we present a tightly integrated query engine that is able to further reduce the volume of intermediate results at run-time. An extensive evaluation shows that our approach improves query execution times by up to two and transferred data volumes by up to three orders of magnitude compared to a naive implementation.",
"Hundreds of public SPARQL endpoints have been deployed on the Web, forming a novel decentralised infrastructure for querying billions of structured facts from a variety of sources on a plethora of topics. But is this infrastructure mature enough to support applications? For 427 public SPARQL endpoints registered on the DataHub, we conduct various experiments to test their maturity. Regarding discoverability, we find that only one-third of endpoints make descriptive meta-data available, making it difficult to locate or learn about their content and capabilities. Regarding interoperability, we find patchy support for established SPARQL features like ORDER BY as well as (understandably) for new SPARQL 1.1 features. Regarding efficiency, we show that the performance of endpoints for generic queries can vary by up to 34 orders of magnitude. Regarding availability, based on a 27-month long monitoring experiment, we show that only 32.2 of public endpoints can be expected to have (monthly) \"two-nines\" uptimes of 99100 .",
"To increase performance, data sharing platforms often make use of clusters of nodes where certain tasks can be executed in parallel. Resource planning and especially deciding how many processors should be chosen to exploit parallel processing is complex in such a setup as increasing the number of processors does not always improve runtime due to communication overhead. Instead, there is usually an optimum number of processors for which using more or fewer processors leads to less efficient runtimes. In this paper, we present a cost model based on widely used statistics (VoiD) and show how to compute the optimum number of processors that should be used to evaluate a particular SPARQL query over a particular configuration and RDF dataset. Our first experiments show the general applicability of our approach but also how shortcomings in the used statistics limit the potential of optimization.",
"Typical approaches for querying structured Web Data collect (crawl) and pre-process (index) large amounts of data in a central data repository before allowing for query answering. However, this time-consuming pre-processing phase however leverages the benefits of Linked Data -- where structured data is accessible live and up-to-date at distributed Web resources that may change constantly -- only to a limited degree, as query results can never be current. An ideal query answering system for Linked Data should return current answers in a reasonable amount of time, even on corpora as large as the Web. Query processors evaluating queries directly on the live sources require knowledge of the contents of data sources. In this paper, we develop and evaluate an approximate index structure summarising graph-structured content of sources adhering to Linked Data principles, provide an algorithm for answering conjunctive queries over Linked Data on theWeb exploiting the source summary, and evaluate the system using synthetically generated queries. The experimental results show that our lightweight index structure enables complete and up-to-date query results over Linked Data, while keeping the overhead for querying low and providing a satisfying source ranking at no additional cost."
]
} |
1705.06135 | 2616586468 | Answering queries over a federation of SPARQL endpoints requires combining data from more than one data source. Optimizing queries in such scenarios is particularly challenging not only because of (i) the large variety of possible query execution plans that correctly answer the query but also because (ii) there is only limited access to statistics about schema and instance data of remote sources. To overcome these challenges, most federated query engines rely on heuristics to reduce the space of possible query execution plans or on dynamic programming strategies to produce optimal plans. Nevertheless, these plans may still exhibit a high number of intermediate results or high execution times because of heuristics and inaccurate cost estimations. In this paper, we present Odyssey, an approach that uses statistics that allow for a more accurate cost estimation for federated queries and therefore enables Odyssey to produce better query execution plans. Our experimental results show that Odyssey produces query execution plans that are better in terms of data transfer and execution time than state-of-the-art optimizers. Our experiments using the FedBench benchmark show execution time gains of at least 25 times on average. | Characteristic sets (CS) @cite_18 @cite_19 aim at solving this problem in centralized systems by capturing statistics about sets of entities having the same set of properties. This information can then be used to accurately estimate the cardinality and join ordering of star-shaped queries. Typically, any set of joined triple patterns in a query can be divided into connected star-shaped subqueries. Subqueries in combination with the predicate that links them, define a characteristic pair (CP) @cite_12 @cite_2 . Statistics about such CPs can then be used to estimate the selectivity of two star-shaped subqueries. Such cardinality estimations can be combined with dynamic programming on a reduced space of alternative query plans. Whereas existing work on CSs and CPs were developed for centralized environments, this paper proposes a solution generalizing these principles for federated environments. | {
"cite_N": [
"@cite_19",
"@cite_18",
"@cite_12",
"@cite_2"
],
"mid": [
"2100387739",
"155517400",
"2396012111",
"2615404014"
],
"abstract": [
"Accurate cardinality estimates are essential for a successful query optimization. This is not only true for relational DBMSs but also for RDF stores. An RDF database consists of a set of triples and, hence, can be seen as a relational database with a single table with three attributes. This makes RDF rather special in that queries typically contain many self joins. We show that relational DBMSs are not well-prepared to perform cardinality estimation in this context. Further, there are hardly any special cardinality estimation methods for RDF databases. To overcome this lack of appropriate cardinality estimation methods, we introduce characteristic sets together with new cardinality estimation methods based upon them. We then show experimentally that the new methods are-in the RDF context-highly superior to the estimation methods employed by commercial DBMSs and by the open-source RDF store RDF-3X.",
"The rapid growth of RDF data in RDF knowledge bases calls for efficient query processing techniques. This paper focuses on the star-style SPARQL join queries, which is very common when users want to search information of entities from RDF knowledge bases. We observe that the computational cost of such queries mainly comes from loading a large portion of predicate-ahead indexes. We therefore propose to partition the whole RDF knowledge bases based on the schema of individual entities, so that only entities of similar schemas are allocated into the same cluster. Such a partitioning strategy generates a pruning mechanism that effectively isolate the correlations of partitions and the queries. Consequently, queries are only conducted over a small number of partitions with small predicate-ahead indexes. Experiments over a large real-life RDF data set show the significant performance improvements achieved by our partitioned indexing techniques.",
"The join ordering problem is a fundamental challenge that has to be solved by any query optimizer. Since the high-performance RDF systems are often implemented as triple stores (i.e., they represent RDF data as a single table with three attributes, at least conceptually), the query optimization strategies employed by such systems are often adopted from relational query optimization. In this paper we show that the techniques borrowed from traditional SQL query optimization (such as Dynamic Programming algorithm or greedy heuristics) are not immediately capable of handling large SPARQL queries. We introduce a new join ordering algorithm that performs a SPARQL-tailored query simplification. Furthermore, we present a novel RDF statistical synopsis that accurately estimates cardinalities in large SPARQL queries. Our experiments show that this algorithm is highly superior to the state-of-the-art SPARQL optimization approaches, including the RDF-3X’s original Dynamic Programming strategy.",
"SPARQL query execution in state of the art RDF engines depends on, and is often limited by the underlying storage and indexing schemes. Typically, these systems exhaustively store permutations of the standard three-column triples table. However, even though RDF can give birth to datasets with loosely defined schemas, it is common for an emerging structure to appear in the data. In this paper, we introduce a novel indexing scheme for RDF data, that takes advantage of the inherent structure of triples. To this end, we define the Extended Characteristic Set (ECS), a schema abstraction that classifies triples based on the properties of their subjects and objects, and we discuss methods and algorithms for the identification and extraction of ECSs. We show how these can be used to assist query processing, and we implement axonDB, an RDF storage and querying engine based on ECS indexing. We perform an experimental evaluation on real world and synthetic datasets and observe that axonDB outperforms the competition by a few orders of magnitude."
]
} |
1705.06391 | 2615907746 | Recent several years have witnessed the surge of asynchronous (async-) parallel computing methods due to the extremely big data involved in many modern applications and also the advancement of multi-core machines and computer clusters. In optimization, most works about async-parallel methods are on unconstrained problems or those with block separable constraints. In this paper, we propose an async-parallel method based on block coordinate update (BCU) for solving convex problems with nonseparable linear constraint. Running on a single node, the method becomes a novel randomized primal-dual BCU with adaptive stepsize for multi-block affinely constrained problems. For these problems, Gauss-Seidel cyclic primal-dual BCU needs strong convexity to have convergence. On the contrary, merely assuming convexity, we show that the objective value sequence generated by the proposed algorithm converges in probability to the optimal value and also the constraint residual to zero. In addition, we establish an ergodic @math convergence result, where @math is the number of iterations. Numerical experiments are performed to demonstrate the efficiency of the proposed method and significantly better speed-up performance than its sync-parallel counterpart. | Running on a single node, the proposed async-parallel method reduces to a serial randomized primal-dual BCU. In the literature, various Gauss-Seidel (GS) cyclic BCU methods have been developed for solving separable convex programs with linear constraints. Although a cyclic primal-dual BCU can empirically work well, in general it may diverge @cite_51 @cite_3 . To guarantee convergence, additional assumptions besides convexity must be made, such as strong convexity on part of the objective @cite_35 @cite_49 @cite_8 @cite_47 @cite_57 @cite_23 @cite_27 and orthogonality properties of block matrices in the linear constraint @cite_53 . Without these assumptions, modifications to the algorithm are necessary for convergence, such as further correction step after each cycle of updates @cite_25 @cite_22 , random permutation of all blocks before each cycle of updates @cite_48 , Jacobi-type update @cite_18 @cite_16 that is essentially linearized augmented Lagrange method (ALM), and hybrid Jacobi-GS update @cite_20 @cite_50 @cite_13 . Different from these modifications, our algorithm simply employs randomization in selecting block variable and can perform significantly better than Jacobi-type methods. In addition, convergence is guaranteed with mere convexity assumption and thus better than those results for GS-type methods. | {
"cite_N": [
"@cite_13",
"@cite_35",
"@cite_18",
"@cite_22",
"@cite_8",
"@cite_48",
"@cite_53",
"@cite_3",
"@cite_57",
"@cite_27",
"@cite_49",
"@cite_23",
"@cite_50",
"@cite_47",
"@cite_16",
"@cite_51",
"@cite_25",
"@cite_20"
],
"mid": [
"",
"2122712590",
"2277973662",
"",
"",
"2161823659",
"1992841740",
"",
"",
"",
"",
"",
"",
"",
"",
"1115208862",
"2007822845",
"1982831910"
],
"abstract": [
"",
"We consider the linearly constrained separable convex programming, whose objective function is separable into m individual convex functions without coupled variables. The alternating direction method of multipliers has been well studied in the literature for the special case m=2, while it remains open whether its convergence can be extended to the general case m≥3. This note shows the global convergence of this extension when the involved functions are further assumed to be strongly convex.",
"This paper introduces a parallel and distributed algorithm for solving the following minimization problem with linear constraints: @math minimizef1(x1)+ź+fN(xN)subject toA1x1+ź+ANxN=c,x1źX1,ź,xNźXN,where @math Nź2, @math fi are convex functions, @math Ai are matrices, and @math Xi are feasible sets for variable @math xi. Our algorithm extends the alternating direction method of multipliers (ADMM) and decomposes the original problem into N smaller subproblems and solves them in parallel at each iteration. This paper shows that the classic ADMM can be extended to the N-block Jacobi fashion and preserve convergence in the following two cases: (i) matrices @math Ai are mutually near-orthogonal and have full column-rank, or (ii) proximal terms are added to the N subproblems (but without any assumption on matrices @math Ai). In the latter case, certain proximal terms can let the subproblem be solved in more flexible and efficient ways. We show that @math źxk+1-xkźM2 converges at a rate of o(1 k) where M is a symmetric positive semi-definte matrix. Since the parameters used in the convergence analysis are conservative, we introduce a strategy for automatically tuning the parameters to substantially accelerate our algorithm in practice. We implemented our algorithm (for the case ii above) on Amazon EC2 and tested it on basis pursuit problems with >300 GB of distributed data. This is the first time that successfully solving a compressive sensing problem of such a large scale is reported.",
"",
"",
"The alternating direction method of multipliers (ADMM) is widely used in solving structured convex optimization problems. Despite its success in practice, the convergence of the standard ADMM for minimizing the sum of (N ,(N 3) ) convex functions, whose variables are linked by linear constraints, has remained unclear for a very long time. Recently, (Math Program, doi:10.1007 s10107-014-0826-5, 2014) provided a counter-example showing that the ADMM for (N 3 ) may fail to converge without further conditions. Since the ADMM for (N 3 ) has been very successful when applied to many problems arising from real practice, it is worth further investigating under what kind of sufficient conditions it can be guaranteed to converge. In this paper, we present such sufficient conditions that can guarantee the sublinear convergence rate for the ADMM for (N 3 ). Specifically, we show that if one of the functions is convex (not necessarily strongly convex) and the other N-1 functions are strongly convex, and the penalty parameter lies in a certain region, the ADMM converges with rate O(1 t) in a certain ergodic sense and o(1 t) in a certain non-ergodic sense, where t denotes the number of iterations. As a by-product, we also provide a simple proof for the O(1 t) convergence rate of two-block ADMM in terms of both objective error and constraint violation, without assuming any condition on the penalty parameter and strong convexity on the functions.",
"The alternating direction method of multipliers (ADMM) is now widely used in many fields, and its convergence was proved when two blocks of variables are alternatively updated. It is strongly desirable and practically valuable to extend the ADMM directly to the case of a multi-block convex minimization problem where its objective function is the sum of more than two separable convex functions. However, the convergence of this extension has been missing for a long time--neither an affirmative convergence proof nor an example showing its divergence is known in the literature. In this paper we give a negative answer to this long-standing open question: The direct extension of ADMM is not necessarily convergent. We present a sufficient condition to ensure the convergence of the direct extension of ADMM, and give an example to show its divergence.",
"",
"",
"",
"",
"",
"",
"",
"",
"We focus on the convergence analysis of the extended linearized alternating direction method of multipliers (L-ADMM) for solving convex minimization problems with three or more separable blocks in the objective functions. Previous convergence analysis of the L-ADMM needs to reduce the multi-block convex minimization problems to two blocks by grouping the variables. Moreover, there has been no rate of convergence analysis for the L-ADMM. In this paper, we construct a counter example to show the failure of convergence of the extended L-ADMM. We prove the convergence and establish the sublinear convergence rate of the extended L-ADMM under the assumptions that the proximal gradient step sizes are smaller than certain values, and any two coefficient matrices in linear constraints are orthogonal.",
"We consider the linearly constrained separable convex minimization problem whose objective function is separable into m individual convex functions with nonoverlapping variables. A Douglas–Rachford alternating direction method of multipliers (ADM) has been well studied in the literature for the special case of @math . But the convergence of extending ADM to the general case of @math is still open. In this paper, we show that the straightforward extension of ADM is valid for the general case of @math if it is combined with a Gaussian back substitution procedure. The resulting ADM with Gaussian back substitution is a novel approach towards the extension of ADM from @math to @math , and its algorithmic framework is new in the literature. For the ADM with Gaussian back substitution, we prove its convergence via the analytic framework of contractive-type methods, and we show its numerical efficiency by some application problems.",
"In this paper, we consider conic programming problems whose constraints consist of linear equalities, linear inequalities, a nonpolyhedral cone, and a polyhedral cone. A convenient way for solving this class of problems is to apply the directly extended alternating direction method of multipliers (ADMM) to its dual problem, which has been observed to perform well in numerical computations but may diverge in theory. Ideally, one should find a convergent variant which is at least as efficient as the directly extended ADMM in practice. We achieve this goal by designing a convergent semiproximal ADMM (called sPADMM3c for convenience) for convex programming problems having three separable blocks in the objective function with the third part being linear. At each iteration, the proposed sPADMM3c takes one special block coordinate descent (BCD) cycle with the order @math , instead of the usual @math Gauss--Seidel BCD cycle used in the nonconvergent directly extended 3-block ADMM, for updating the variable blocks. Our numerical experiments demonstrate that the convergent method is at least 20 faster than the directly extended ADMM with unit step-length for the vast majority of about 550 large-scale doubly nonnegative semidefinite programming problems with linear equality and or inequality constraints. This confirms that at least for conic convex programming, one can design a convergent and efficient ADMM with a special BCD cycle of updating the variable blocks."
]
} |
1705.05899 | 2951879426 | As LoRaWAN networks are actively being deployed in the field, it is important to comprehend the limitations of this Low Power Wide Area Network technology. Previous work has raised questions in terms of the scalability and capacity of LoRaWAN networks as the number of end devices grows to hundreds or thousands per gateway. Some works have modeled LoRaWAN networks as pure ALOHA networks, which fails to capture important characteristics such as the capture effect and the effects of interference. Other works provide a more comprehensive model by relying on empirical and stochastic techniques. This work uses a different approach where a LoRa error model is constructed from extensive complex baseband bit error rate simulations and used as an interference model. The error model is combined with the LoRaWAN MAC protocol in an ns-3 module that enables to study multi channel, multi spreading factor, multi gateway, bi-directional LoRaWAN networks with thousands of end devices. Using the lorawan ns-3 module, a scalability analysis of LoRaWAN shows the detrimental impact of downstream traffic on the delivery ratio of confirmed upstream traffic. The analysis shows that increasing gateway density can ameliorate but not eliminate this effect, as stringent duty cycle requirements for gateways continue to limit downstream opportunities. | @cite_5 , Georgiou and Raza provide a stochastic geometry framework for modeling the performance of a single channel LoRa network. Two independent link-outage conditions are studied, one which is related to SNR (i.e. range) and another one which is related to co-spreading factor interference. The authors argue that LoRa networks will inevitably become interference-limited, as end device coverage probability decays exponentially with increasing number of end devices. The authors report that this is mostly caused by co-spreading factor interference and that the low duty cycle and chirp orthogonality found in LoRa do little to mitigate this. Finally, the authors note that the lack of a packet-level software simulation is hindering the study into the performance of LoRa. It would be interesting to combine the authors' modeling of co-spreading factor interference with our ns-3 error model, as in the SINR approach all interference is treated as noise. | {
"cite_N": [
"@cite_5"
],
"mid": [
"2537187694"
],
"abstract": [
"Low power wide area (LPWA) networks are making spectacular progress from design, standardization, to commercialization. At this time of fast-paced adoption, it is of utmost importance to analyze how well these technologies will scale as the number of devices connected to the Internet of Things inevitably grows. In this letter, we provide a stochastic geometry framework for modeling the performance of a single gateway LoRa network, a leading LPWA technology. Our analysis formulates the unique peculiarities of LoRa, including its chirp spread-spectrum modulation technique, regulatory limitations on radio duty cycle, and use of ALOHA protocol on top, all of which are not as common in today’s commercial cellular networks. We show that the coverage probability drops exponentially as the number of end-devices grows due to interfering signals using the same spreading sequence. We conclude that this fundamental limiting factor is perhaps more significant toward LoRa scalability than for instance spectrum restrictions. Our derivations for co-spreading factor interference found in LoRa networks enables rigorous scalability analysis of such networks."
]
} |
1705.05899 | 2951879426 | As LoRaWAN networks are actively being deployed in the field, it is important to comprehend the limitations of this Low Power Wide Area Network technology. Previous work has raised questions in terms of the scalability and capacity of LoRaWAN networks as the number of end devices grows to hundreds or thousands per gateway. Some works have modeled LoRaWAN networks as pure ALOHA networks, which fails to capture important characteristics such as the capture effect and the effects of interference. Other works provide a more comprehensive model by relying on empirical and stochastic techniques. This work uses a different approach where a LoRa error model is constructed from extensive complex baseband bit error rate simulations and used as an interference model. The error model is combined with the LoRaWAN MAC protocol in an ns-3 module that enables to study multi channel, multi spreading factor, multi gateway, bi-directional LoRaWAN networks with thousands of end devices. Using the lorawan ns-3 module, a scalability analysis of LoRaWAN shows the detrimental impact of downstream traffic on the delivery ratio of confirmed upstream traffic. The analysis shows that increasing gateway density can ameliorate but not eliminate this effect, as stringent duty cycle requirements for gateways continue to limit downstream opportunities. | The work of @cite_6 studies the limit on the number of transmitters supported by a LoRa system based on an empirical model. The authors performed practical experiments that quantify communication range and capture effect of LoRa transmissions. These findings were used to build a purpose-built simulator, LoRaSim, with the goal of studying the scalability of LoRa networks. The authors conclude that LoRa networks can scale quite well if they use dynamic transmissions parameter selection and or multiple sinks. Our study confirms that multiple sinks drastically improve scalability, even though we use a very different approach for modeling interference. Furthermore, our study goes deeper into modeling LoRaWAN as the LoRaWAN MAC layer is modeled and the impact of confirmed messages and downstream traffic is studied. | {
"cite_N": [
"@cite_6"
],
"mid": [
"2525332113"
],
"abstract": [
"New Internet of Things (IoT) technologies such as Long Range (LoRa) are emerging which enable power efficient wireless communication over very long distances. Devices typically communicate directly to a sink node which removes the need of constructing and maintaining a complex multi-hop network. Given the fact that a wide area is covered and that all devices communicate directly to a few sink nodes a large number of nodes have to share the communication medium. LoRa provides for this reason a range of communication options (centre frequency, spreading factor, bandwidth, coding rates) from which a transmitter can choose. Many combination settings are orthogonal and provide simultaneous collision free communications. Nevertheless, there is a limit regarding the number of transmitters a LoRa system can support. In this paper we investigate the capacity limits of LoRa networks. Using experiments we develop models describing LoRa communication behaviour. We use these models to parameterise a LoRa simulation to study scalability. Our experiments show that a typical smart city deployment can support 120 nodes per 3.8 ha, which is not sufficient for future IoT deployments. LoRa networks can scale quite well, however, if they use dynamic communication parameter selection and or multiple sinks."
]
} |
1705.06025 | 2616703740 | Fingerprinting based WLAN indoor positioning system (FWIPS) provides a promising indoor positioning solution to meet the growing interests for indoor location-based services (e.g., indoor way finding or geo-fencing). FWIPS is preferred because it requires no additional infrastructure for deploying an FWIPS and achieving the position estimation by reusing the available WLAN and mobile devices, and capable of providing absolute position estimation. For fingerprinting based positioning (FbP), a model is created to provide reference values of observable features (e.g., signal strength from access point (AP)) as a function of location during offline stage. One widely applied method to build a complete and an accurate reference database (i.e. radio map (RM)) for FWIPS is carrying out a site survey throughout the region of interest (RoI). Along the site survey, the readings of received signal strength (RSS) from all visible APs at each reference point (RP) are collected. This site survey, however, is time-consuming and labor-intensive, especially in the case that the RoI is large (e.g., an airport or a big mall). This bottleneck hinders the wide commercial applications of FWIPS (e.g., proximity promotions in a shopping center). To diminish the cost of site survey, we propose a probabilistic model, which combines fingerprinting based positioning (FbP) and RM generation based on stochastic variational Bayesian inference (SVBI). This SVBI based position and RSS estimation has three properties: i) being able to predict the distribution of the estimated position and RSS, ii) treating each observation of RSS at each RP as an example to learn for FbP and RM generation instead of using the whole RM as an example, and iii) requiring only one time training of the SVBI model for both localization and RSS estimation. These benefits make it outperforms the previous proposed approaches. | There are few publications addressing simultaneous fbp and rm generation. @cite_16 and @cite_30 employ compressive sensing with @math regularization and manifold alignment with geometry perturbation to achieve both fbp and rm generation. In @cite_13 , propose a method using nn with backpropagation to realize joint position estimation and rm generation. However, the intrinsic discrepancy of the dimensionality between rss readings and the coordinates requires to train the position estimation and rm generation models separately, i.e. there are two different training processes: one for mapping the rss to the coordinate space to achieve fbp , and another one for transforming the coordinates to rss space for rm generation. | {
"cite_N": [
"@cite_30",
"@cite_16",
"@cite_13"
],
"mid": [
"2343041401",
"2011502582",
""
],
"abstract": [
"The Received Signal Strength (RSS) based fingerprinting approaches for indoor localization pose a need for updating the fingerprint databases due to dynamic nature of the indoor environment. This process is hectic and time-consuming when the size of the indoor area is large. The semi-supervised approaches reduce this workload and achieve good accuracy around 15 percent of the fingerprinting load but the performance is severely degraded if it is reduced below this level. We propose an indoor localization framework that uses unsupervised manifold alignment. It requires only 1 percent of the fingerprinting load, some crowd sourced readings, and plan coordinates of the indoor area. The 1 percent fingerprinting load is used only in perturbing the local geometries of the plan coordinates. The proposed framework achieves less than 5 m mean localization error, which is considerably better than semi-supervised approaches at very small amount of fingerprinting load. In addition, the few location estimations together with few fingerprints help to estimate the complete radio map of the indoor environment. The estimation of radio map does not demand extra workload rather it employs the already available information from the proposed indoor localization framework. The testing results for radio map estimation show almost 50 percent performance improvement by using this information as compared to using only fingerprints.",
"The recent growing interest for indoor Location-Based Services (LBSs) has created a need for more accurate and real-time indoor positioning solutions. The sparse nature of location finding makes the theory of Compressive Sensing (CS) desirable for accurate indoor positioning using Received Signal Strength (RSS) from Wireless Local Area Network (WLAN) Access Points (APs). We propose an accurate RSS-based indoor positioning system using the theory of compressive sensing, which is a method to recover sparse signals from a small number of noisy measurements by solving an 1-minimization problem. Our location estimator consists of a coarse localizer, where the RSS is compared to a number of clusters to detect in which cluster the node is located, followed by a fine localization step, using the theory of compressive sensing, to further refine the location estimation. We have investigated different coarse localization schemes and AP selection approaches to increase the accuracy. We also show that the CS theory can be used to reconstruct the RSS radio map from measurements at only a small number of fingerprints, reducing the number of measurements significantly. We have implemented the proposed system on a WiFi-integrated mobile device and have evaluated the performance. Experimental results indicate that the proposed system leads to substantial improvement on localization accuracy and complexity over the widely used traditional fingerprinting methods.",
""
]
} |
1705.05994 | 2770703741 | We propose the Variational Shape Learner (VSL), a generative model that learns the underlying structure of voxelized 3D shapes in an unsupervised fashion. Through the use of skip-connections, our model can successfully learn and infer a latent, hierarchical representation of objects. Furthermore, realistic 3D objects can be easily generated by sampling the VSL's latent probabilistic manifold. We show that our generative model can be trained end-to-end from 2D images to perform single image 3D model retrieval. Experiments show, both quantitatively and qualitatively, the improved generalization of our proposed model over a range of tasks, performing better or comparable to various state-of-the-art alternatives. | Research in learning probabilistic generative models has also benefited from the advances made by artificial neural networks. Generative Adversarial Networks (GANs), proposed in @cite_22 and Variational auto-encoders (VAEs), proposed in @cite_43 @cite_28 , are some of the most popular and important frameworks that have emerged from improvements in generative modeling. Successful adaptation of these frameworks range from a focus in natural language and speech processing @cite_15 @cite_5 to realistic image synthesis @cite_25 @cite_37 @cite_33 , yielding promising, positive results. Nevertheless, very little work, outside of @cite_17 @cite_34 @cite_9 , has focused on modeling 3D objects, where generative architectures can be used to learn probabilistic embeddings. The model proposed in this paper will offer another step towards constructing powerful probabilistic generative models of 3D structures. | {
"cite_N": [
"@cite_37",
"@cite_22",
"@cite_33",
"@cite_28",
"@cite_9",
"@cite_43",
"@cite_5",
"@cite_15",
"@cite_34",
"@cite_25",
"@cite_17"
],
"mid": [
"2173520492",
"2099471712",
"2951326654",
"1909320841",
"2469266052",
"",
"2734812719",
"2950067852",
"2335364074",
"1850742715",
"2949551726"
],
"abstract": [
"In recent years, supervised learning with convolutional networks (CNNs) has seen huge adoption in computer vision applications. Comparatively, unsupervised learning with CNNs has received less attention. In this work we hope to help bridge the gap between the success of CNNs for supervised learning and unsupervised learning. We introduce a class of CNNs called deep convolutional generative adversarial networks (DCGANs), that have certain architectural constraints, and demonstrate that they are a strong candidate for unsupervised learning. Training on various image datasets, we show convincing evidence that our deep convolutional adversarial pair learns a hierarchy of representations from object parts to scenes in both the generator and discriminator. Additionally, we use the learned features for novel tasks - demonstrating their applicability as general image representations.",
"We propose a new framework for estimating generative models via an adversarial process, in which we simultaneously train two models: a generative model G that captures the data distribution, and a discriminative model D that estimates the probability that a sample came from the training data rather than G. The training procedure for G is to maximize the probability of D making a mistake. This framework corresponds to a minimax two-player game. In the space of arbitrary functions G and D, a unique solution exists, with G recovering the training data distribution and D equal to ½ everywhere. In the case where G and D are defined by multilayer perceptrons, the entire system can be trained with backpropagation. There is no need for any Markov chains or unrolled approximate inference networks during either training or generation of samples. Experiments demonstrate the potential of the framework through qualitative and quantitative evaluation of the generated samples.",
"A novel variational autoencoder is developed to model images, as well as associated labels or captions. The Deep Generative Deconvolutional Network (DGDN) is used as a decoder of the latent image features, and a deep Convolutional Neural Network (CNN) is used as an image encoder; the CNN is used to approximate a distribution for the latent DGDN features code. The latent code is also linked to generative models for labels (Bayesian support vector machine) or captions (recurrent neural network). When predicting a label caption for a new image at test, averaging is performed across the distribution of latent codes; this is computationally efficient as a consequence of the learned CNN-based encoder. Since the framework is capable of modeling the image in the presence absence of associated labels captions, a new semi-supervised setting is manifested for CNN learning with images; the framework even allows unsupervised CNN learning, based on images alone.",
"We marry ideas from deep neural networks and approximate Bayesian inference to derive a generalised class of deep, directed generative models, endowed with a new algorithm for scalable inference and learning. Our algorithm introduces a recognition model to represent approximate posterior distributions, and that acts as a stochastic encoder of the data. We develop stochastic back-propagation -- rules for back-propagation through stochastic variables -- and use this to develop an algorithm that allows for joint optimisation of the parameters of both the generative and recognition model. We demonstrate on several real-world data sets that the model generates realistic samples, provides accurate imputations of missing data and is a useful tool for high-dimensional data visualisation.",
"A key goal of computer vision is to recover the underlying 3D structure from 2D observations of the world. In this paper we learn strong deep generative models of 3D structures, and recover these structures from 3D and 2D images via probabilistic inference. We demonstrate high-quality samples and report log-likelihoods on several datasets, including ShapeNet [2], and establish the first benchmarks in the literature. We also show how these models and their inference networks can be trained end-to-end from 2D images. This demonstrates for the first time the feasibility of learning to infer 3D representations of the world in a purely unsupervised manner.",
"",
"Advances in neural variational inference have facilitated the learning of powerful directed graphical models with continuous latent variables, such as variational autoencoders. The hope is that such models will learn to represent rich, multi-modal latent factors in real-world data, such as natural language text. However, current models often assume simplistic priors on the latent variables - such as the uni-modal Gaussian distribution - which are incapable of representing complex latent factors efficiently. To overcome this restriction, we propose the simple, but highly flexible, piecewise constant distribution. This distribution has the capacity to represent an exponential number of modes of a latent target distribution, while remaining mathematically tractable. Our results demonstrate that incorporating this new latent distribution into different models yields substantial improvements in natural language processing tasks such as document modeling and natural language generation for dialogue.",
"In this paper, we explore the inclusion of latent random variables into the dynamic hidden state of a recurrent neural network (RNN) by combining elements of the variational autoencoder. We argue that through the use of high-level latent random variables, the variational RNN (VRNN)1 can model the kind of variability observed in highly structured sequential data such as natural speech. We empirically evaluate the proposed model against related sequential models on four speech datasets and one handwriting dataset. Our results show the important roles that latent random variables can play in the RNN dynamic hidden state.",
"What is a good vector representation of an object? We believe that it should be generative in 3D, in the sense that it can produce new 3D objects; as well as be predictable from 2D, in the sense that it can be perceived from 2D images. We propose a novel architecture, called the TL-embedding network, to learn an embedding space with these properties. The network consists of two components: (a) an autoencoder that ensures the representation is generative; and (b) a convolutional network that ensures the representation is predictable. This enables tackling a number of tasks including voxel prediction from 2D images and 3D model retrieval. Extensive experimental analysis demonstrates the usefulness and versatility of this embedding.",
"This paper introduces the Deep Recurrent Attentive Writer (DRAW) neural network architecture for image generation. DRAW networks combine a novel spatial attention mechanism that mimics the foveation of the human eye, with a sequential variational auto-encoding framework that allows for the iterative construction of complex images. The system substantially improves on the state of the art for generative models on MNIST, and, when trained on the Street View House Numbers dataset, it generates images that cannot be distinguished from real data with the naked eye.",
"We study the problem of 3D object generation. We propose a novel framework, namely 3D Generative Adversarial Network (3D-GAN), which generates 3D objects from a probabilistic space by leveraging recent advances in volumetric convolutional networks and generative adversarial nets. The benefits of our model are three-fold: first, the use of an adversarial criterion, instead of traditional heuristic criteria, enables the generator to capture object structure implicitly and to synthesize high-quality 3D objects; second, the generator establishes a mapping from a low-dimensional probabilistic space to the space of 3D objects, so that we can sample objects without a reference image or CAD models, and explore the 3D object manifold; third, the adversarial discriminator provides a powerful 3D shape descriptor which, learned without supervision, has wide applications in 3D object recognition. Experiments demonstrate that our method generates high-quality 3D objects, and our unsupervisedly learned features achieve impressive performance on 3D object recognition, comparable with those of supervised learning methods."
]
} |
1705.05994 | 2770703741 | We propose the Variational Shape Learner (VSL), a generative model that learns the underlying structure of voxelized 3D shapes in an unsupervised fashion. Through the use of skip-connections, our model can successfully learn and infer a latent, hierarchical representation of objects. Furthermore, realistic 3D objects can be easily generated by sampling the VSL's latent probabilistic manifold. We show that our generative model can be trained end-to-end from 2D images to perform single image 3D model retrieval. Experiments show, both quantitatively and qualitatively, the improved generalization of our proposed model over a range of tasks, performing better or comparable to various state-of-the-art alternatives. | One study, amidst the rise of neural network-based approaches to 3D object recognition, most relevant to this paper is that of @cite_27 , which presented promising results and a useful benchmark for 3D model recognition: ModelNet. Following this key study, researchers have tried applying 3D ConvNets @cite_44 @cite_46 @cite_2 @cite_40 , autoencoders @cite_26 @cite_39 @cite_31 @cite_49 , and a variety of probabilistic neural generative models @cite_17 @cite_9 to the problem of 3D model recognition, with each study progressively advancing state-of-the-art. | {
"cite_N": [
"@cite_31",
"@cite_26",
"@cite_9",
"@cite_39",
"@cite_44",
"@cite_40",
"@cite_27",
"@cite_49",
"@cite_2",
"@cite_46",
"@cite_17"
],
"mid": [
"",
"1922697897",
"2469266052",
"",
"2211722331",
"",
"2951755740",
"",
"",
"2951911415",
"2949551726"
],
"abstract": [
"",
"Complex geometric structural variations of 3D model usually pose great challenges in 3D shape matching and retrieval. In this paper, we propose a high-level shape feature learning scheme to extract features that are insensitive to deformations via a novel discriminative deep auto-encoder. First, a multiscale shape distribution is developed for use as input to the auto-encoder. Then, by imposing the Fisher discrimination criterion on the neurons in the hidden layer, we developed a novel discriminative deep auto-encoder for shape feature learning. Finally, the neurons in the hidden layers from multiple discriminative auto-encoders are concatenated to form a shape descriptor for 3D shape matching and retrieval. The proposed method is evaluated on the representative datasets that contain 3D models with large geometric variations, i.e., Mcgill and SHREC'10 ShapeGoogle datasets. Experimental results on the benchmark datasets demonstrate the effectiveness of the proposed method for 3D shape matching and retrieval.",
"A key goal of computer vision is to recover the underlying 3D structure from 2D observations of the world. In this paper we learn strong deep generative models of 3D structures, and recover these structures from 3D and 2D images via probabilistic inference. We demonstrate high-quality samples and report log-likelihoods on several datasets, including ShapeNet [2], and establish the first benchmarks in the literature. We also show how these models and their inference networks can be trained end-to-end from 2D images. This demonstrates for the first time the feasibility of learning to infer 3D representations of the world in a purely unsupervised manner.",
"",
"Robust object recognition is a crucial skill for robots operating autonomously in real world environments. Range sensors such as LiDAR and RGBD cameras are increasingly found in modern robotic systems, providing a rich source of 3D information that can aid in this task. However, many current systems do not fully utilize this information and have trouble efficiently dealing with large amounts of point cloud data. In this paper, we propose VoxNet, an architecture to tackle this problem by integrating a volumetric Occupancy Grid representation with a supervised 3D Convolutional Neural Network (3D CNN). We evaluate our approach on publicly available benchmarks using LiDAR, RGBD, and CAD data. VoxNet achieves accuracy beyond the state of the art while labeling hundreds of instances per second.",
"",
"3D shape is a crucial but heavily underutilized cue in today's computer vision systems, mostly due to the lack of a good generic shape representation. With the recent availability of inexpensive 2.5D depth sensors (e.g. Microsoft Kinect), it is becoming increasingly important to have a powerful 3D shape representation in the loop. Apart from category recognition, recovering full 3D shapes from view-based 2.5D depth maps is also a critical part of visual understanding. To this end, we propose to represent a geometric 3D shape as a probability distribution of binary variables on a 3D voxel grid, using a Convolutional Deep Belief Network. Our model, 3D ShapeNets, learns the distribution of complex 3D shapes across different object categories and arbitrary poses from raw CAD data, and discovers hierarchical compositional part representations automatically. It naturally supports joint object recognition and shape completion from 2.5D depth maps, and it enables active object recognition through view planning. To train our 3D deep learning model, we construct ModelNet -- a large-scale 3D CAD model dataset. Extensive experiments show that our 3D deep representation enables significant performance improvement over the-state-of-the-arts in a variety of tasks.",
"",
"",
"Inspired by the recent success of methods that employ shape priors to achieve robust 3D reconstructions, we propose a novel recurrent neural network architecture that we call the 3D Recurrent Reconstruction Neural Network (3D-R2N2). The network learns a mapping from images of objects to their underlying 3D shapes from a large collection of synthetic data. Our network takes in one or more images of an object instance from arbitrary viewpoints and outputs a reconstruction of the object in the form of a 3D occupancy grid. Unlike most of the previous works, our network does not require any image annotations or object class labels for training or testing. Our extensive experimental analysis shows that our reconstruction framework i) outperforms the state-of-the-art methods for single view reconstruction, and ii) enables the 3D reconstruction of objects in situations when traditional SFM SLAM methods fail (because of lack of texture and or wide baseline).",
"We study the problem of 3D object generation. We propose a novel framework, namely 3D Generative Adversarial Network (3D-GAN), which generates 3D objects from a probabilistic space by leveraging recent advances in volumetric convolutional networks and generative adversarial nets. The benefits of our model are three-fold: first, the use of an adversarial criterion, instead of traditional heuristic criteria, enables the generator to capture object structure implicitly and to synthesize high-quality 3D objects; second, the generator establishes a mapping from a low-dimensional probabilistic space to the space of 3D objects, so that we can sample objects without a reference image or CAD models, and explore the 3D object manifold; third, the adversarial discriminator provides a powerful 3D shape descriptor which, learned without supervision, has wide applications in 3D object recognition. Experiments demonstrate that our method generates high-quality 3D objects, and our unsupervisedly learned features achieve impressive performance on 3D object recognition, comparable with those of supervised learning methods."
]
} |
1705.05994 | 2770703741 | We propose the Variational Shape Learner (VSL), a generative model that learns the underlying structure of voxelized 3D shapes in an unsupervised fashion. Through the use of skip-connections, our model can successfully learn and infer a latent, hierarchical representation of objects. Furthermore, realistic 3D objects can be easily generated by sampling the VSL's latent probabilistic manifold. We show that our generative model can be trained end-to-end from 2D images to perform single image 3D model retrieval. Experiments show, both quantitatively and qualitatively, the improved generalization of our proposed model over a range of tasks, performing better or comparable to various state-of-the-art alternatives. | With respect to 3D object generation from 2D images, commonly used methods can be roughly grouped into two categories: 3D voxel prediction @cite_27 @cite_17 @cite_34 @cite_9 @cite_46 @cite_50 and mesh-based methods @cite_19 @cite_1 . The 3D-R2N2 model @cite_46 represents a more recent approach to the task, which involves training a recurrent neural network to predict 3D voxels from one or more 2D images. @cite_9 also takes a recurrent network-based approach, but receives a depth image as input rather than normal 2D images. The learnable stereo system @cite_3 processes one or more camera views and camera pose information to produce compelling 3D object samples. | {
"cite_N": [
"@cite_9",
"@cite_1",
"@cite_3",
"@cite_19",
"@cite_27",
"@cite_50",
"@cite_46",
"@cite_34",
"@cite_17"
],
"mid": [
"2469266052",
"2013483851",
"2748099314",
"2104697781",
"2951755740",
"2606840594",
"2951911415",
"2335364074",
"2949551726"
],
"abstract": [
"A key goal of computer vision is to recover the underlying 3D structure from 2D observations of the world. In this paper we learn strong deep generative models of 3D structures, and recover these structures from 3D and 2D images via probabilistic inference. We demonstrate high-quality samples and report log-likelihoods on several datasets, including ShapeNet [2], and establish the first benchmarks in the literature. We also show how these models and their inference networks can be trained end-to-end from 2D images. This demonstrates for the first time the feasibility of learning to infer 3D representations of the world in a purely unsupervised manner.",
"This article proposes a variational multi-view stereo vision method based on meshes for recovering 3D scenes (shape and radiance) from images. Our method is based on generative models and minimizes the reprojection error (difference between the observed images and the images synthesized from the reconstruction). Our contributions are twofold. 1) For the first time, we rigorously compute the gradient of the reprojection error for non smooth surfaces defined by discrete triangular meshes. The gradient correctly takes into account the visibility changes that occur when a surface moves; this forces the contours generated by the reconstructed surface to perfectly match with the apparent contours in the input images. 2) We propose an original modification of the Lambertian model to take into account deviations from the constant brightness assumption without explicitly modelling the reflectance properties of the scene or other photometric phenomena involved by the camera model. Our method is thus able to recover the shape and the diffuse radiance of non Lambertian scenes.",
"We present a learnt system for multi-view stereopsis. In contrast to recent learning based methods for 3D reconstruction, we leverage the underlying 3D geometry of the problem through feature projection and unprojection along viewing rays. By formulating these operations in a differentiable manner, we are able to learn the system end-to-end for the task of metric 3D reconstruction. End-to-end learning allows us to jointly reason about shape priors while conforming geometric constraints, enabling reconstruction from much fewer images (even a single image) than required by classical approaches as well as completion of unseen surfaces. We thoroughly evaluate our approach on the ShapeNet dataset and demonstrate the benefits over classical approaches as well as recent learning based methods.",
"This paper addresses the problem of image-based surface reconstruction. The main contribution is the computation of the exact derivative of the reprojection error functional. This allows its rigorous minimization via gradient descent surface evolution. The main difficulty has been to correctly take into account the visibility changes that occur when the surface moves. A geometric and analytical study of these changes is presented and used for the computation of derivative. Our analysis shows the strong influence that the movement of the contour generators has on the reprojection error. As a consequence, during the proper minimization of the reprojection error, the contour generators of the surface are automatically moved to their correct location in the images. Therefore, current methods adding additional silhouettes or apparent contour constraints to ensure this alignment can now be understood and justified by a single criterion: the reprojection error.",
"3D shape is a crucial but heavily underutilized cue in today's computer vision systems, mostly due to the lack of a good generic shape representation. With the recent availability of inexpensive 2.5D depth sensors (e.g. Microsoft Kinect), it is becoming increasingly important to have a powerful 3D shape representation in the loop. Apart from category recognition, recovering full 3D shapes from view-based 2.5D depth maps is also a critical part of visual understanding. To this end, we propose to represent a geometric 3D shape as a probability distribution of binary variables on a 3D voxel grid, using a Convolutional Deep Belief Network. Our model, 3D ShapeNets, learns the distribution of complex 3D shapes across different object categories and arbitrary poses from raw CAD data, and discovers hierarchical compositional part representations automatically. It naturally supports joint object recognition and shape completion from 2.5D depth maps, and it enables active object recognition through view planning. To train our 3D deep learning model, we construct ModelNet -- a large-scale 3D CAD model dataset. Extensive experiments show that our 3D deep representation enables significant performance improvement over the-state-of-the-arts in a variety of tasks.",
"Recently, Convolutional Neural Networks have shown promising results for 3D geometry prediction. They can make predictions from very little input data such as a single color image. A major limitation of such approaches is that they only predict a coarse resolution voxel grid, which does not capture the surface of the objects well. We propose a general framework, called hierarchical surface prediction (HSP), which facilitates prediction of high resolution voxel grids. The main insight is that it is sufficient to predict high resolution voxels around the predicted surfaces. The exterior and interior of the objects can be represented with coarse resolution voxels. Our approach is not dependent on a specific input type. We show results for geometry prediction from color images, depth images and shape completion from partial voxel grids. Our analysis shows that our high resolution predictions are more accurate than low resolution predictions.",
"Inspired by the recent success of methods that employ shape priors to achieve robust 3D reconstructions, we propose a novel recurrent neural network architecture that we call the 3D Recurrent Reconstruction Neural Network (3D-R2N2). The network learns a mapping from images of objects to their underlying 3D shapes from a large collection of synthetic data. Our network takes in one or more images of an object instance from arbitrary viewpoints and outputs a reconstruction of the object in the form of a 3D occupancy grid. Unlike most of the previous works, our network does not require any image annotations or object class labels for training or testing. Our extensive experimental analysis shows that our reconstruction framework i) outperforms the state-of-the-art methods for single view reconstruction, and ii) enables the 3D reconstruction of objects in situations when traditional SFM SLAM methods fail (because of lack of texture and or wide baseline).",
"What is a good vector representation of an object? We believe that it should be generative in 3D, in the sense that it can produce new 3D objects; as well as be predictable from 2D, in the sense that it can be perceived from 2D images. We propose a novel architecture, called the TL-embedding network, to learn an embedding space with these properties. The network consists of two components: (a) an autoencoder that ensures the representation is generative; and (b) a convolutional network that ensures the representation is predictable. This enables tackling a number of tasks including voxel prediction from 2D images and 3D model retrieval. Extensive experimental analysis demonstrates the usefulness and versatility of this embedding.",
"We study the problem of 3D object generation. We propose a novel framework, namely 3D Generative Adversarial Network (3D-GAN), which generates 3D objects from a probabilistic space by leveraging recent advances in volumetric convolutional networks and generative adversarial nets. The benefits of our model are three-fold: first, the use of an adversarial criterion, instead of traditional heuristic criteria, enables the generator to capture object structure implicitly and to synthesize high-quality 3D objects; second, the generator establishes a mapping from a low-dimensional probabilistic space to the space of 3D objects, so that we can sample objects without a reference image or CAD models, and explore the 3D object manifold; third, the adversarial discriminator provides a powerful 3D shape descriptor which, learned without supervision, has wide applications in 3D object recognition. Experiments demonstrate that our method generates high-quality 3D objects, and our unsupervisedly learned features achieve impressive performance on 3D object recognition, comparable with those of supervised learning methods."
]
} |
1705.05994 | 2770703741 | We propose the Variational Shape Learner (VSL), a generative model that learns the underlying structure of voxelized 3D shapes in an unsupervised fashion. Through the use of skip-connections, our model can successfully learn and infer a latent, hierarchical representation of objects. Furthermore, realistic 3D objects can be easily generated by sampling the VSL's latent probabilistic manifold. We show that our generative model can be trained end-to-end from 2D images to perform single image 3D model retrieval. Experiments show, both quantitatively and qualitatively, the improved generalization of our proposed model over a range of tasks, performing better or comparable to various state-of-the-art alternatives. | Many of the above methods require multiple images and or additional human-provided information. Some approaches have attempted to minimize human involvement by developing weakly-supervised schemes, making use of image silhouettes to conduct 3D object reconstruction @cite_41 @cite_35 . Of the few unsupervised neural-based approaches that exist, the T-L network @cite_34 is quite important, which combines a convolutional autoencoder with an image regressor to encode a unified vector representation of a given 2D image. However, one fundamental issue with the T-L Network is its three-phase training procedure, since jointly training the system components proves to be too difficult. The 3D-GAN @cite_17 offers a way to train 3D object models probabilistically, employing an adversarial learning scheme. However, GANs are notoriously difficult to train @cite_13 , often due to ill-designed loss functions and the higher chance of zero gradients. | {
"cite_N": [
"@cite_35",
"@cite_41",
"@cite_34",
"@cite_13",
"@cite_17"
],
"mid": [
"2770907521",
"2950701417",
"2335364074",
"2581485081",
"2949551726"
],
"abstract": [
"The objective of this paper is 3D shape understanding from single and multiple images. To this end, we introduce a new deep-learning architecture and loss function, SilNet, that can handle multiple views in an order-agnostic manner. The architecture is fully convolutional, and for training we use a proxy task of silhouette prediction, rather than directly learning a mapping from 2D images to 3D shape as has been the target in most recent work. We demonstrate that with the SilNet architecture there is generalisation over the number of views -- for example, SilNet trained on 2 views can be used with 3 or 4 views at test-time; and performance improves with more views. We introduce two new synthetics datasets: a blobby object dataset useful for pre-training, and a challenging and realistic sculpture dataset; and demonstrate on these datasets that SilNet has indeed learnt 3D shape. Finally, we show that SilNet exceeds the state of the art on the ShapeNet benchmark dataset, and use SilNet to generate novel views of the sculpture dataset.",
"Understanding the 3D world is a fundamental problem in computer vision. However, learning a good representation of 3D objects is still an open problem due to the high dimensionality of the data and many factors of variation involved. In this work, we investigate the task of single-view 3D object reconstruction from a learning agent's perspective. We formulate the learning process as an interaction between 3D and 2D representations and propose an encoder-decoder network with a novel projection loss defined by the perspective transformation. More importantly, the projection loss enables the unsupervised learning using 2D observation without explicit 3D supervision. We demonstrate the ability of the model in generating 3D volume from a single 2D image with three sets of experiments: (1) learning from single-class objects; (2) learning from multi-class objects and (3) testing on novel object classes. Results show superior performance and better generalization ability for 3D object reconstruction when the projection loss is involved.",
"What is a good vector representation of an object? We believe that it should be generative in 3D, in the sense that it can produce new 3D objects; as well as be predictable from 2D, in the sense that it can be perceived from 2D images. We propose a novel architecture, called the TL-embedding network, to learn an embedding space with these properties. The network consists of two components: (a) an autoencoder that ensures the representation is generative; and (b) a convolutional network that ensures the representation is predictable. This enables tackling a number of tasks including voxel prediction from 2D images and 3D model retrieval. Extensive experimental analysis demonstrates the usefulness and versatility of this embedding.",
"The goal of this paper is not to introduce a single algorithm or method, but to make theoretical steps towards fully understanding the training dynamics of generative adversarial networks. In order to substantiate our theoretical analysis, we perform targeted experiments to verify our assumptions, illustrate our claims, and quantify the phenomena. This paper is divided into three sections. The first section introduces the problem at hand. The second section is dedicated to studying and proving rigorously the problems including instability and saturation that arize when training generative adversarial networks. The third section examines a practical and theoretically grounded direction towards solving these problems, while introducing new tools to study them.",
"We study the problem of 3D object generation. We propose a novel framework, namely 3D Generative Adversarial Network (3D-GAN), which generates 3D objects from a probabilistic space by leveraging recent advances in volumetric convolutional networks and generative adversarial nets. The benefits of our model are three-fold: first, the use of an adversarial criterion, instead of traditional heuristic criteria, enables the generator to capture object structure implicitly and to synthesize high-quality 3D objects; second, the generator establishes a mapping from a low-dimensional probabilistic space to the space of 3D objects, so that we can sample objects without a reference image or CAD models, and explore the 3D object manifold; third, the adversarial discriminator provides a powerful 3D shape descriptor which, learned without supervision, has wide applications in 3D object recognition. Experiments demonstrate that our method generates high-quality 3D objects, and our unsupervisedly learned features achieve impressive performance on 3D object recognition, comparable with those of supervised learning methods."
]
} |
1705.05908 | 2949077439 | We study the evolution of long-lived controversial debates as manifested on Twitter from 2011 to 2016. Specifically, we explore how the structure of interactions and content of discussion varies with the level of collective attention, as evidenced by the number of users discussing a topic. Spikes in the volume of users typically correspond to external events that increase the public attention on the topic -- as, for instance, discussions about gun control' often erupt after a mass shooting. This work is the first to study the dynamic evolution of polarized online debates at such scale. By employing a wide array of network and content analysis measures, we find consistent evidence that increased collective attention is associated with increased network polarization and network concentration within each side of the debate; and overall more uniform lexicon usage across all users. | Several studies have looked at how networks evolve, and proposed models of network formation @cite_6 @cite_11 . Densification over time is a pattern often observed @cite_6 , i.e., social networks gain more edges as the number of nodes grows. A change in the scaling behavior of the degree distribution has also been observed @cite_3 . offer a comprehensive review. Most of these studies focus on social networks, and in particular, on the friendship relationship. In our work, we are interested in studying an network, which has markedly different characteristics. | {
"cite_N": [
"@cite_3",
"@cite_6",
"@cite_11"
],
"mid": [
"2121761994",
"2111708605",
"2151078464"
],
"abstract": [
"Social networking services are a fast-growing business in the Internet. However, it is unknown if online relationships and their growth patterns are the same as in real-life social networks. In this paper, we compare the structures of three online social networking services: Cyworld, MySpace, and orkut, each with more than 10 million users, respectively. We have access to complete data of Cyworld's ilchon (friend) relationships and analyze its degree distribution, clustering property, degree correlation, and evolution over time. We also use Cyworld data to evaluate the validity of snowball sampling method, which we use to crawl and obtain partial network topologies of MySpace and orkut. Cyworld, the oldest of the three, demonstrates a changing scaling behavior over time in degree distribution. The latest Cyworld data's degree distribution exhibits a multi-scaling behavior, while those of MySpace and orkut have simple scaling behaviors with different exponents. Very interestingly, each of the two e ponents corresponds to the different segments in Cyworld's degree distribution. Certain online social networking services encourage online activities that cannot be easily copied in real life; we show that they deviate from close-knit online social networks which show a similar degree correlation pattern to real-life social networks.",
"How do real graphs evolve over time? What are \"normal\" growth patterns in social, technological, and information networks? Many studies have discovered patterns in static graphs, identifying properties in a single snapshot of a large network, or in a very small number of snapshots; these include heavy tails for in- and out-degree distributions, communities, small-world phenomena, and others. However, given the lack of information about network evolution over long periods, it has been hard to convert these findings into statements about trends over time.Here we study a wide range of real graphs, and we observe some surprising phenomena. First, most of these graphs densify over time, with the number of edges growing super-linearly in the number of nodes. Second, the average distance between nodes often shrinks over time, in contrast to the conventional wisdom that such distance parameters should increase slowly as a function of the number of nodes (like O(log n) or O(log(log n)).Existing graph generation models do not exhibit these types of behavior, even at a qualitative level. We provide a new graph generator, based on a \"forest fire\" spreading process, that has a simple, intuitive justification, requires very few parameters (like the \"flammability\" of nodes), and produces graphs exhibiting the full range of properties observed both in prior work and in the present study.",
"We present a detailed study of network evolution by analyzing four large online social networks with full temporal information about node and edge arrivals. For the first time at such a large scale, we study individual node arrival and edge creation processes that collectively lead to macroscopic properties of networks. Using a methodology based on the maximum-likelihood principle, we investigate a wide variety of network formation strategies, and show that edge locality plays a critical role in evolution of networks. Our findings supplement earlier network models based on the inherently non-local preferential attachment. Based on our observations, we develop a complete model of network evolution, where nodes arrive at a prespecified rate and select their lifetimes. Each node then independently initiates edges according to a \"gap\" process, selecting a destination for each edge according to a simple triangle-closing model free of any parameters. We show analytically that the combination of the gap distribution with the node lifetime leads to a power law out-degree distribution that accurately reflects the true network in all four cases. Finally, we give model parameter settings that allow automatic evolution and generation of realistic synthetic networks of arbitrary scale."
]
} |
1705.06031 | 2614905167 | Accuracy is one of the basic principles of journalism. However, it is increasingly hard to manage due to the diversity of news media. Some editors of online news tend to use catchy headlines which trick readers into clicking. These headlines are either ambiguous or misleading, degrading the reading experience of the audience. Thus, identifying inaccurate news headlines is a task worth studying. Previous work names these headlines "clickbaits" and mainly focus on the features extracted from the headlines, which limits the performance since the consistency between headlines and news bodies is underappreciated. In this paper, we clearly redefine the problem and identify ambiguous and misleading headlines separately. We utilize class sequential rules to exploit structure information when detecting ambiguous headlines. For the identification of misleading headlines, we extract features based on the congruence between headlines and bodies. To make use of the large unlabeled data set, we apply a co-training method and gain an increase in performance. The experiment results show the effectiveness of our methods. Then we use our classifiers to detect inaccurate headlines crawled from different sources and conduct a data analysis. | In Communication and Psychology, there have been studies on the accuracy of news headlines since decades ago. Marquez divided news headlines into three types, namely accurate, ambiguous and misleading ones, and proposed specific definitions, which are subsequently used in our classification. Ecker analysed the effect of misinformation in news headlines and showed that such headlines lead to misconception in readers. Several properties and structures of clickbait headlines were dug out manually @cite_2 @cite_3 @cite_15 , providing an entry point for preliminary automatic identification of clickbaits. | {
"cite_N": [
"@cite_15",
"@cite_3",
"@cite_2"
],
"mid": [
"2248267741",
"2143482575",
"2046592476"
],
"abstract": [
"Tabloid journalism is often criticized for its propensity for exaggeration, sensationalization, scare-mongering, and otherwise producing misleading and low quality news. As the news has moved online, a new form of tabloidization has emerged: ?clickbaiting.? ?Clickbait? refers to ?content whose main purpose is to attract attention and encourage visitors to click on a link to a particular web page? [?clickbait,? n.d.] and has been implicated in the rapid spread of rumor and misinformation online. This paper examines potential methods for the automatic detection of clickbait as a form of deception. Methods for recognizing both textual and non-textual clickbaiting cues are surveyed, leading to the suggestion that a hybrid approach may yield best results.",
"This article explores the application of metaphors in news headlines with a view to interrogating their potential for coercion. Coercion in news discourse is understood as a strategic deployment of pragma-linguistic devices, including metaphors, to foreground the representations of socio-political reality that are compatible with the interests of the news outlet rather than those that inform public debate. It is argued that coercion can be exposed through systematic discourse analysis. Methodologically, the study aims to integrate the cognitive and pragmatic approaches to metaphor in regarding it as both a conceptual building block of news representations and a strategic framing device in news discourse. A quantitative and qualitative analysis of a sample of metaphors excerpted from a corpus of 400 most-read headlines from one of the most visited English-language newspaper sites The Daily Mail is conducted to illustrate such coercive applications of metaphor as simplification, imaging, animalization, confronta tion, (de)legitimization, emotionalization, and dramatization. In the course of the analysis it is demonstrated how certain ideologically-biased representations can be coerced through figurative language.",
"This article sets out a framework for a language-oriented analysis of sensationalism in news media. Sensationalism is understood here as a discourse strategy of ‘packaging’ information in news headlines in such a way that news items are presented as more interesting, extraordinary and relevant than might be the case. Unlike previous content analyses of sensational coverage, this study demonstrates how sensationalism is instantiated through specific illocutions, semantic macrostructures, narrative formulas, evaluation parameters, and interpersonal and textual devices. Examples are drawn from a corpus of headlines of the ‘most read’ articles in the online outlet of the British mid-market tabloid Daily Mail compiled in early 2012. Sensationalized instances are identified through surveys and focus group discussions and subsequently analyzed quantitatively and qualitatively. The study is located within the context of media scholarship on news values and current trends in journalism."
]
} |
1705.05787 | 2614376571 | We propose formulations for learning features for Offline Signature Verification.A novel method that uses knowledge of forgeries from a subset of users is proposed.Learned features are used to train classifiers for other users (without forgeries).Experiments on GPDS-960 show a large improvement in state-of-the-art.Results in other 3 datasets show that the features generalize without fine-tuning. Verifying the identity of a person using handwritten signatures is challenging in the presence of skilled forgeries, where a forger has access to a persons signature and deliberately attempt to imitate it. In offline (static) signature verification, the dynamic information of the signature writing process is lost, and it is difficult to design good feature extractors that can distinguish genuine signatures and skilled forgeries. This reflects in a relatively poor performance, with verification errors around 7 in the best systems in the literature. To address both the difficulty of obtaining good features, as well as improve system performance, we propose learning the representations from signature images, in a Writer-Independent format, using Convolutional Neural Networks. In particular, we propose a novel formulation of the problem that includes knowledge of skilled forgeries from a subset of users in the feature learning process, that aims to capture visual cues that distinguish genuine signatures and forgeries regardless of the user. Extensive experiments were conducted on four datasets: GPDS, MCYT, CEDAR and Brazilian PUC-PR datasets. On GPDS-160, we obtained a large improvement in state-of-the-art performance, achieving 1.72 Equal Error Rate, compared to 6.97 in the literature. We also verified that the features generalize beyond the GPDS dataset, surpassing the state-of-the-art performance in the other datasets, without requiring the representation to be fine-tuned to each particular dataset. | The area of automatic Offline Signature Verification has been researched at least since the decade of 1970. Over the years, the problem has been addressed from many different perspectives, as summarized by @cite_9 , @cite_44 and @cite_47 . | {
"cite_N": [
"@cite_44",
"@cite_9",
"@cite_47"
],
"mid": [
"2131991463",
"2062219129",
""
],
"abstract": [
"In recent years, along with the extraordinary diffusion of the Internet and a growing need for personal verification in many daily applications, automatic signature verification is being considered with renewed interest. This paper presents the state of the art in automatic signature verification. It addresses the most valuable results obtained so far and highlights the most profitable directions of research to date. It includes a comprehensive bibliography of more than 300 selected references as an aid for researchers working in the field.",
"Abstract This paper presents a survey of the literature on automatic signature verification and writer identification by computer, and an overview of achievements in static and dynamic approaches to solving these problems, with a special focus on preprocessing techniques, feature extraction methods, comparison processes and performance evaluation. In addition, for each type of approache special attention is given to requirement analysis, human factors, practical application environments, and appropriate definitions and terminology. Throughout the paper, new research directions are suggested.",
""
]
} |
1705.05787 | 2614376571 | We propose formulations for learning features for Offline Signature Verification.A novel method that uses knowledge of forgeries from a subset of users is proposed.Learned features are used to train classifiers for other users (without forgeries).Experiments on GPDS-960 show a large improvement in state-of-the-art.Results in other 3 datasets show that the features generalize without fine-tuning. Verifying the identity of a person using handwritten signatures is challenging in the presence of skilled forgeries, where a forger has access to a persons signature and deliberately attempt to imitate it. In offline (static) signature verification, the dynamic information of the signature writing process is lost, and it is difficult to design good feature extractors that can distinguish genuine signatures and skilled forgeries. This reflects in a relatively poor performance, with verification errors around 7 in the best systems in the literature. To address both the difficulty of obtaining good features, as well as improve system performance, we propose learning the representations from signature images, in a Writer-Independent format, using Convolutional Neural Networks. In particular, we propose a novel formulation of the problem that includes knowledge of skilled forgeries from a subset of users in the feature learning process, that aims to capture visual cues that distinguish genuine signatures and forgeries regardless of the user. Extensive experiments were conducted on four datasets: GPDS, MCYT, CEDAR and Brazilian PUC-PR datasets. On GPDS-160, we obtained a large improvement in state-of-the-art performance, achieving 1.72 Equal Error Rate, compared to 6.97 in the literature. We also verified that the features generalize beyond the GPDS dataset, surpassing the state-of-the-art performance in the other datasets, without requiring the representation to be fine-tuned to each particular dataset. | There are mainly two approaches for building offline signature verification systems. The most common approach is to design Writer-Dependent classifiers. In this scenario, a training set is constructed for each user of the system, consisting of genuine signatures as positive examples and genuine signatures from other users (random forgeries) as negative samples. A binary classifier is then trained on this dataset, resulting in one model for each user. This approach has shown to work well for the task, but since it requires one model to be trained for each user, complexity increases as more users are enrolled. An alternative is Writer-Independent classification. In this case, a single model is trained for all users, by training a classifier in a dissimilarity space @cite_27 , @cite_19 . The inputs for classification are dissimilarity vectors, that represent the difference between the features of a query signature, and the features of a template signature (a genuine signature of the user). In spite of the reduced complexity, Writer-Independent systems often perform worse, and the best results in standard benchmarks are obtained with Writer-Dependent systems. | {
"cite_N": [
"@cite_19",
"@cite_27"
],
"mid": [
"2075069329",
"1964159865"
],
"abstract": [
"Standard signature verification (SV) systems are writer-dependent (WD), where a specific classifier is designed for each individual. It is inconvenient to ask a user to provide enough number of signature samples to design his WD classifier. In practice, very few samples are collected and inaccurate classifiers maybe produced. To overcome this, writer-independent (WI) systems are introduced. A global classifier is designed using a development database, prior to enrolling users to the system. For these systems, signature templates are needed for verification, and the template databases can be compromised. Moreover, state-of-the-art WI and WD systems provide enhanced accuracy through information fusion at either feature, score or decision levels, but they increase computational complexity. In this study, a hybrid WI-WD system is proposed, as a compromise of the two approaches. When a user is enrolled to the system, a WI classifier is used to verify his queries. During operation, user samples are collected and adapt the WI classifier to his signatures. Once adapted, the resulting WD classifier replaces the WI classifier for this user. Simulations on the Brazilian and the GPDS signature databases indicate that the proposed hybrid system provides comparative accuracy as complex WI and WD systems, while decreases the classification complexity.",
"In this work we address two important issues of off-line signature verification. The first one regards feature extraction. We introduce a new graphometric feature set that considers the curvature of the most important segments, perceptually speaking, of the signature. The idea is to simulate the shape of the signature by using Bezier curves and then extract features from these curves. The second important aspect is the use of an ensemble of classifiers based on graphometric features to improve the reliability of the classification, hence reducing the false acceptance. The ensemble was built using a standard genetic algorithm and different fitness functions were assessed to drive the search. Two different scenarios were considered in our experiments. In the former, we assume that only genuine signatures and random forgeries are available to guide the search. In the latter, on the other hand, we assume that simple and simulated forgeries also are available during the optimization of the ensemble. The pool of base classifiers is trained using only genuine signatures and random forgeries. Thorough experiments were conduct on a database composed of 100 writers and the results compare favorably."
]
} |
1705.05787 | 2614376571 | We propose formulations for learning features for Offline Signature Verification.A novel method that uses knowledge of forgeries from a subset of users is proposed.Learned features are used to train classifiers for other users (without forgeries).Experiments on GPDS-960 show a large improvement in state-of-the-art.Results in other 3 datasets show that the features generalize without fine-tuning. Verifying the identity of a person using handwritten signatures is challenging in the presence of skilled forgeries, where a forger has access to a persons signature and deliberately attempt to imitate it. In offline (static) signature verification, the dynamic information of the signature writing process is lost, and it is difficult to design good feature extractors that can distinguish genuine signatures and skilled forgeries. This reflects in a relatively poor performance, with verification errors around 7 in the best systems in the literature. To address both the difficulty of obtaining good features, as well as improve system performance, we propose learning the representations from signature images, in a Writer-Independent format, using Convolutional Neural Networks. In particular, we propose a novel formulation of the problem that includes knowledge of skilled forgeries from a subset of users in the feature learning process, that aims to capture visual cues that distinguish genuine signatures and forgeries regardless of the user. Extensive experiments were conducted on four datasets: GPDS, MCYT, CEDAR and Brazilian PUC-PR datasets. On GPDS-160, we obtained a large improvement in state-of-the-art performance, achieving 1.72 Equal Error Rate, compared to 6.97 in the literature. We also verified that the features generalize beyond the GPDS dataset, surpassing the state-of-the-art performance in the other datasets, without requiring the representation to be fine-tuned to each particular dataset. | A large variety of feature extractors have been investigated for this problem, from simple geometric descriptors @cite_26 , @cite_0 , descriptors inspired in graphology and graphometry @cite_30 , directional-based descriptors such as HOG @cite_48 and D-PDF @cite_35 , @cite_20 , @cite_19 , descriptors based on interest-point, such as SIFT @cite_48 , to texture descriptors, such as Local Binary Patterns (LBP) @cite_48 and Gray-Level Co-occurrence Matrix (GLCM) @cite_14 . These features are commonly extracted locally from the signature images, by dividing the image in a grid and computing descriptors for each cell (either in Cartesian or polar coordinates). | {
"cite_N": [
"@cite_30",
"@cite_35",
"@cite_26",
"@cite_14",
"@cite_48",
"@cite_0",
"@cite_19",
"@cite_20"
],
"mid": [
"2529690827",
"1512670244",
"1821815814",
"2053219667",
"2280957996",
"61605974",
"2075069329",
"2150428256"
],
"abstract": [
"",
"The first stage of a complete automatic handwritten signature verification system (AHSVS) is described in this paper. Since only random forgeries are taken into account in this first stage of decision, the directional probability density function (PDF) which is related to the overall shape of the handwritten signature has been taken into account as feature vector. Experimental results show that using both directional PDFs and the completely connected feedforward neural network classifier are valuable to build the first stage of a complete AHSVS. >",
"This paper deals with the detection of freehand forgeries of signatures on bank checks. The detection process makes use of size ratio and slant features derived from Eden's kinematic stroke model for handwriting, which was modified to make it applicable to prewritten material. The features are measured for a real signature by a process involving automatic thresholding, to extract the signature from the background; analysis of projections, to segment the signature into vertical zones; detection of tall letters, to segment it into horizontal zones; and identification of the tall letters with respect to the (assumed known) spelling of the signature. Statistical assumptions are made regarding the expected variation in feature values among different writers and for a single writer. Tests on a small data base led to verification of these assumptions and to successful forgery detection.",
"We present an offline signature verification system using three different pseudo-dynamic features, two different classifier training approaches and two datasets. One of the most difficult problems of off-line signature verification is that the signature is just a static image while losing a lot of useful dynamic information. Three separate pseudo-dynamic features based on gray level: local binary pattern (LBP), gray level co-occurrence matrix (GLCM) and histogram of oriented gradients (HOG) are used. The classification is performed using writer-dependent Support Vector Machine (SVMs) classifiers and Global Real Adaboost method, where two different approaches to train the classifier. In the first mode, each SVM is trained with the feature vectors obtained from the reference signatures of the corresponding user and those random forgeries for each signer while the global Adaboost classifier is trained using genuine and random forgery signatures of signers that are excluded from the test set. The fusion of all features achieves the best result of 7.66 and 9.94 equal error rate in GPDS while 7.55 and 11.55 equal error rate in CSD respectively.",
"A state-of-art offline signature verification system is presented.Complementary approaches and features are fused at multiple levels.Useful local binary patterns are selected for offline signature verification.In-depth analysis of the effects of alignment for registering signatures is done.A new feature is proposed from SIFT keypoint alignment. Offline signature verification is a task that benefits from matching both the global shape and local details; as such, it is particularly suitable to a fusion approach. We present a system that uses a score-level fusion of complementary classifiers that use different local features (histogram of oriented gradients, local binary patterns and scale invariant feature transform descriptors), where each classifier uses a feature-level fusion to represent local features at coarse-to-fine levels. For classifiers, two different approaches are investigated, namely global and user-dependent classifiers. User-dependent classifiers are trained separately for each user, to learn to differentiate that users genuine signatures from other signatures; while a single global classifier is trained with difference vectors of query and reference signatures of all users in the training set, to learn the importance of different types of dissimilarities.The fusion of all classifiers achieves a state-of-the-art performance with 6.97 equal error rate in skilled forgery tests using the public GPDS-160 signature database. The proposed system does not require skilled forgeries of the enrolling user, which is essential for real life applications.",
"The objective of this work is to present an off-line signature verification system. It is basically divided into three parts. The first one describes a pre-processing process, a segmentation process and a feature extraction process, in which the main aim is to obtain the maximum performance quality of the process of verification of random falsifications, in the false acceptance and false rejection concept. The second presents a learning process based on HMM, where the aim is obtaining the best model, that is, one that is capable of representing each writer's signature, absorbing yet at the same time discriminating, at most the intra-personal variation and the interpersonal variation. A third and last part, presents a signature verification process that uses the models generated by the learning process without using any prior knowledge of test data, in other words, using an automatic derivation process of the decision thresholds.",
"Standard signature verification (SV) systems are writer-dependent (WD), where a specific classifier is designed for each individual. It is inconvenient to ask a user to provide enough number of signature samples to design his WD classifier. In practice, very few samples are collected and inaccurate classifiers maybe produced. To overcome this, writer-independent (WI) systems are introduced. A global classifier is designed using a development database, prior to enrolling users to the system. For these systems, signature templates are needed for verification, and the template databases can be compromised. Moreover, state-of-the-art WI and WD systems provide enhanced accuracy through information fusion at either feature, score or decision levels, but they increase computational complexity. In this study, a hybrid WI-WD system is proposed, as a compromise of the two approaches. When a user is enrolled to the system, a WI classifier is used to verify his queries. During operation, user samples are collected and adapt the WI classifier to his signatures. Once adapted, the resulting WD classifier replaces the WI classifier for this user. Simulations on the Brazilian and the GPDS signature databases indicate that the proposed hybrid system provides comparative accuracy as complex WI and WD systems, while decreases the classification complexity.",
"Some of the fundamental problems faced in the design of signature verification (SV) systems include the potentially large number of input features and users, the limited number of reference signatures for training, the high intra-personal variability among signatures, and the lack of forgeries as counterexamples. In this paper, a new approach for feature selection is proposed for writer-independent (WI) off-line SV. First, one or more preexisting techniques are employed to extract features at different scales. Multiple feature extraction increases the diversity of information produced from signature images, allowing to produce signature representations that mitigate intra-personal variability. Dichotomy transformation is then applied in the resulting feature space to allow for WI classification. This alleviates the challenges of designing off-line SV systems with a limited number of reference signatures from a large number of users. Finally, boosting feature selection is used to design low-cost classifiers that automatically select relevant features while training. Using this global WI feature selection approach allows to explore and select from large feature sets based on knowledge of a population of users. Experiments performed with real-world SV data comprised of random, simple, and skilled forgeries indicate that the proposed approach provides a high level of performance when extended shadow code and directional probability density function features are extracted at multiple scales. Comparing simulation results to those of off-line SV systems found in literature confirms the viability of the new approach, even when few reference signatures are available. Moreover, it provides an efficient framework for designing a wide range of biometric systems from limited samples with few or no counterexamples, but where new training samples emerge during operations."
]
} |
1705.05787 | 2614376571 | We propose formulations for learning features for Offline Signature Verification.A novel method that uses knowledge of forgeries from a subset of users is proposed.Learned features are used to train classifiers for other users (without forgeries).Experiments on GPDS-960 show a large improvement in state-of-the-art.Results in other 3 datasets show that the features generalize without fine-tuning. Verifying the identity of a person using handwritten signatures is challenging in the presence of skilled forgeries, where a forger has access to a persons signature and deliberately attempt to imitate it. In offline (static) signature verification, the dynamic information of the signature writing process is lost, and it is difficult to design good feature extractors that can distinguish genuine signatures and skilled forgeries. This reflects in a relatively poor performance, with verification errors around 7 in the best systems in the literature. To address both the difficulty of obtaining good features, as well as improve system performance, we propose learning the representations from signature images, in a Writer-Independent format, using Convolutional Neural Networks. In particular, we propose a novel formulation of the problem that includes knowledge of skilled forgeries from a subset of users in the feature learning process, that aims to capture visual cues that distinguish genuine signatures and forgeries regardless of the user. Extensive experiments were conducted on four datasets: GPDS, MCYT, CEDAR and Brazilian PUC-PR datasets. On GPDS-160, we obtained a large improvement in state-of-the-art performance, achieving 1.72 Equal Error Rate, compared to 6.97 in the literature. We also verified that the features generalize beyond the GPDS dataset, surpassing the state-of-the-art performance in the other datasets, without requiring the representation to be fine-tuned to each particular dataset. | Methods to learn features from data have not yet been widely explored for offline signature verification. @cite_16 used Restricted Boltzmann Machines (RBMs) to learn features from signature images. However, in this work they only showed the visual appearance of the weights, and did not test the features for classification. Khalajzadeh @cite_29 used Convolutional Neural Networks (CNNs) for signature verification on a dataset of Persian signatures, but only considered the classification between different users (e.g. detecting random forgeries), and did not considered skilled forgeries. @cite_15 proposed a solution using deep neural networks for Multitask Metric Learning. In their work, a distance metric between pairs of signatures is learned. Contrary to our work, the authors used handcrafted feature extractors (LBP in the experiments with the GPDS dataset), while in our work the inputs to the system are the signature themselves (pixel intensities), and the feature representation is learned. In a similar vein to our work, Eskander @cite_19 presented a hybrid Writer-Independent Writer-Dependent solution, using a Development dataset for feature selection, followed by training WD classifiers using the selected features. However, in the present work we use a Development dataset for feature learning instead of feature selection. | {
"cite_N": [
"@cite_19",
"@cite_29",
"@cite_16",
"@cite_15"
],
"mid": [
"2075069329",
"1601291380",
"39287937",
"2410931070"
],
"abstract": [
"Standard signature verification (SV) systems are writer-dependent (WD), where a specific classifier is designed for each individual. It is inconvenient to ask a user to provide enough number of signature samples to design his WD classifier. In practice, very few samples are collected and inaccurate classifiers maybe produced. To overcome this, writer-independent (WI) systems are introduced. A global classifier is designed using a development database, prior to enrolling users to the system. For these systems, signature templates are needed for verification, and the template databases can be compromised. Moreover, state-of-the-art WI and WD systems provide enhanced accuracy through information fusion at either feature, score or decision levels, but they increase computational complexity. In this study, a hybrid WI-WD system is proposed, as a compromise of the two approaches. When a user is enrolled to the system, a WI classifier is used to verify his queries. During operation, user samples are collected and adapt the WI classifier to his signatures. Once adapted, the resulting WD classifier replaces the WI classifier for this user. Simulations on the Brazilian and the GPDS signature databases indicate that the proposed hybrid system provides comparative accuracy as complex WI and WD systems, while decreases the classification complexity.",
"",
"Reliable identification and verification of off-line handwritten signatures from images is a difficult problem with many practical applications. This task is a difficult vision problem within the field of biometrics because a signature may change depending on psychological factors of the individual. Motivated by advances in brain science which describe how objects are represented in the visual cortex, advanced research on deep neural networks has been shown to work reliably on large image data sets. In this paper, we present a deep learning model for off-line handwritten signature recognition which is able to extract high-level representations. We also propose a two-step hybrid model for signature identification and verification improving the misclassification rate in the well-known GPDS database.",
"A new deep multitask learning based metric learning method is proposed.An appropriate architecture is proposed for offline signature verification.The knowledge of other signers' samples is transfered for better accuracy.The proposed method outperforms SVM and Discriminative Deep Metric Learning.It outperforms the accuracy of other published reports. Display Omitted This paper presents a novel classification method, Deep Multitask Metric Learning (DMML), for offline signature verification. Unlike existing methods that to verify questioned signatures of an individual merely consider the training samples of that class, DMML uses the knowledge from the similarities and dissimilarities between the genuine and forged samples of other classes too. To this end, using the idea of multitask and transfer learning, DMML train a distance metric for each class together with other classes simultaneously. DMML has a structure with a shared layer acting as a writer-independent approach, that is followed by separated layers which learn writer-dependent factors. We compare the proposed method against SVM, writer-dependent and writer-independent Discriminative Deep Metric Learning method on four offline signature datasets (UTSig, MCYT-75, GPDSsynthetic, and GPDS960GraySignatures) using Histogram of Oriented Gradients (HOG) and Discrete Radon Transform (DRT) features. Results of our experiments show that DMML achieves better performance compared to other methods in verifying genuine signatures, skilled and random forgeries."
]
} |
1705.05742 | 2724395316 | The availability of large scale event data with time stamps has given rise to dynamically evolving knowledge graphs that contain temporal information for each edge. Reasoning over time in such dynamic knowledge graphs is not yet well understood. To this end, we present Know-Evolve, a novel deep evolutionary knowledge network that learns non-linearly evolving entity representations over time. The occurrence of a fact (edge) is modeled as a multivariate point process whose intensity function is modulated by the score for that fact computed based on the learned entity embeddings. We demonstrate significantly improved performance over various relational learning approaches on two large scale real-world datasets. Further, our method effectively predicts occurrence or recurrence time of a fact which is novel compared to prior reasoning approaches in multi-relational setting. | Translation Based Models. uses two relation-specific matrices to project subject and object entities and computes @math distance to score a fact between two entity vectors. @cite_7 proposed TransE model that computes score as a distance between relation-specific translations of entity embeddings. improved TransE by allowing entities to have distributed representations on relation specific hyperplane where distance between them is computed. TransR extends this model to use separate semantic spaces for entities and relations and does translation in the relationship space. | {
"cite_N": [
"@cite_7"
],
"mid": [
"2127795553"
],
"abstract": [
"We consider the problem of embedding entities and relationships of multi-relational data in low-dimensional vector spaces. Our objective is to propose a canonical model which is easy to train, contains a reduced number of parameters and can scale up to very large databases. Hence, we propose TransE, a method which models relationships by interpreting them as translations operating on the low-dimensional embeddings of the entities. Despite its simplicity, this assumption proves to be powerful since extensive experiments show that TransE significantly outperforms state-of-the-art methods in link prediction on two knowledge bases. Besides, it can be successfully trained on a large scale data set with 1M entities, 25k relationships and more than 17M training samples."
]
} |
1705.05720 | 2614919369 | Knowledge bases (KBs) have attracted increasing attention due to its great success in various areas, such as Web and mobile search.Existing KBs are restricted to objective factual knowledge, such as city population or fruit shape, whereas,subjective knowledge, such as big city, which is commonly mentioned in Web and mobile queries, has been neglected. Subjective knowledge differs from objective knowledge in that it has no documented or observed ground truth. Instead, the truth relies on people's dominant opinion. Thus, we can use the crowdsourcing technique to get opinion from the crowd. In our work, we propose a system, called crowdsourced subjective knowledge acquisition (CoSKA),for subjective knowledge acquisition powered by crowdsourcing and existing KBs. The acquired knowledge can be used to enrich existing KBs in the subjective dimension which bridges the gap between existing objective knowledge and subjective queries.The main challenge of CoSKA is the conflict between large scale knowledge facts and limited crowdsourcing resource. To address this challenge, in this work, we define knowledge inference rules and then select the seed knowledge judiciously for crowdsourcing to maximize the inference power under the resource constraint. Our experimental results on real knowledge base and crowdsourcing platform verify the effectiveness of CoSKA system. | Subjective knowledge acquisition is closely related to works that associating properties with entities. Some works have been conducted for commonsense knowledge acquisition @cite_19 @cite_29 @cite_0 @cite_26 . WebChild @cite_2 presents a method for automatically constructing a large commonsense knowledge base, it contains triples that connect nouns with adjectives via fine-grained relations. Entitytagger @cite_8 , presented by , automatically associate descriptive phrases, referred to as etags (entity tags), to each entity. Instead of , these works focus on the less controversial and more objective properties, which is not related to obtaining . The most similar work is , which mines the dominant opinion on the web content of whether a subjective property applies to a type. However, they does not consider to use the existing information in knowledge base and resorting to the crowd for subjective knowledge acquisition. | {
"cite_N": [
"@cite_26",
"@cite_8",
"@cite_29",
"@cite_0",
"@cite_19",
"@cite_2"
],
"mid": [
"1862719289",
"2097371712",
"2398622378",
"2792460139",
"2016089260",
"2083897630"
],
"abstract": [
"Applications are increasingly expected to make smart decisions based on what humans consider basic commonsense. An often overlooked but essential form of commonsense involves comparisons, e.g. the fact that bears are typically more dangerous than dogs, that tables are heavier than chairs, or that ice is colder than water. In this paper, we first rely on open information extraction methods to obtain large amounts of comparisons from the Web. We then develop a joint optimization model for cleaning and disambiguating this knowledge with respect to WordNet. This model relies on integer linear programming and semantic coherence scores. Experiments show that our model outperforms strong baselines and allows us to obtain a large knowledge base of disambiguated commonsense assertions.",
"We consider the problem of entity tagging: given one or more named entities from a specific domain, the goal is to automatically associate descriptive phrases, referred to as etags (entity tags), to each entity. Consider a product catalog containing product names and possibly short descriptions. For a product in the catalog, say Ricoh G600 Digital Camera, we want to associate etags such as \"water resistant\", \"rugged\" and \"outdoor\" to it, even though its name or description does not mention those phrases. Entity tagging can enable more effective search over entities. We propose to leverage signals in web documents to perform such tagging. We develop techniques to perform such tagging in a domain independent manner while ensuring high precision and high recall.",
"Knowledge acquisition is the essential process of extracting and encoding knowledge, both domain specific and commonsense, to be used in intelligent systems. While many large knowledge bases have been constructed, none is close to complete. This paper presents an approach to improving a knowledge base efficiently under resource constraints. Using a guiding knowledge base, questions are generated from a weak form of similarity-based inference given the glossary mapping between two knowledge bases. The candidate questions are prioritized in terms of the concept coverage of the target knowledge. Experiments were conducted to find questions to grow the Chinese ConceptNet using the English ConceptNet as a guide. The results were evaluated by online users to verify that 94.17 of the questions and 85.77 of the answers are good. In addition, the answers collected in a six-week period showed consistent improvement to a 36.33 increase in concept coverage of the Chinese commonsense knowledge base against the English ConceptNet.",
"",
"ConceptNet is a freely available commonsense knowledge base and natural-language-processing tool-kit which supports many practical textual-reasoning tasks over real-world documents including topic-gisting, analogy-making, and other context oriented inferences. The knowledge base is a semantic network presently consisting of over 1.6 million assertions of commonsense knowledge encompassing the spatial, physical, social, temporal, and psychological aspects of everyday life. ConceptNet is generated automatically from the 700 000 sentences of the Open Mind Common Sense Project — a World Wide Web based collaboration with over 14 000 authors.",
"This paper presents a method for automatically constructing a large commonsense knowledge base, called WebChild, from Web contents. WebChild contains triples that connect nouns with adjectives via fine-grained relations like hasShape, hasTaste, evokesEmotion, etc. The arguments of these assertions, nouns and adjectives, are disambiguated by mapping them onto their proper WordNet senses. Our method is based on semi-supervised Label Propagation over graphs of noisy candidate assertions. We automatically derive seeds from WordNet and by pattern matching from Web text collections. The Label Propagation algorithm provides us with domain sets and range sets for 19 different relations, and with confidence-ranked assertions between WordNet senses. Large-scale experiments demonstrate the high accuracy (more than 80 percent) and coverage (more than four million fine grained disambiguated assertions) of WebChild."
]
} |
1705.05720 | 2614919369 | Knowledge bases (KBs) have attracted increasing attention due to its great success in various areas, such as Web and mobile search.Existing KBs are restricted to objective factual knowledge, such as city population or fruit shape, whereas,subjective knowledge, such as big city, which is commonly mentioned in Web and mobile queries, has been neglected. Subjective knowledge differs from objective knowledge in that it has no documented or observed ground truth. Instead, the truth relies on people's dominant opinion. Thus, we can use the crowdsourcing technique to get opinion from the crowd. In our work, we propose a system, called crowdsourced subjective knowledge acquisition (CoSKA),for subjective knowledge acquisition powered by crowdsourcing and existing KBs. The acquired knowledge can be used to enrich existing KBs in the subjective dimension which bridges the gap between existing objective knowledge and subjective queries.The main challenge of CoSKA is the conflict between large scale knowledge facts and limited crowdsourcing resource. To address this challenge, in this work, we define knowledge inference rules and then select the seed knowledge judiciously for crowdsourcing to maximize the inference power under the resource constraint. Our experimental results on real knowledge base and crowdsourcing platform verify the effectiveness of CoSKA system. | Knowledge base enrichment, completion and population have been widely studied. There are two mainstreams: internal methods, which use only the knowledge contained in the knowledge base to predict missing information @cite_11 @cite_16 ; external methods,which use sources of knowledge such as text corpora or other knowledge base to add new knowledge facts @cite_17 @cite_12 @cite_28 @cite_24 . However, these works are limited to add and neglect . Moreover, they do not consider to make use of a natural source of knowledge, the crowd, to complete enrich the existing knowledge base. | {
"cite_N": [
"@cite_11",
"@cite_28",
"@cite_24",
"@cite_16",
"@cite_12",
"@cite_17"
],
"mid": [
"1758759808",
"2119465010",
"",
"2151502664",
"2141657856",
"2128407051"
],
"abstract": [
"While the realization of the SemanticWeb as once envisioned by Tim Berners-Lee remains in a distant future, the Web of Data has already become a reality. Billions of RDF statements on the Internet, facts about a variety of different domains, are ready to be used by semantic applications. Some of these applications, however, crucially hinge on the availability of expressive schemas suitable for logical inference that yields non-trivial conclusions. In this paper, we present a statistical approach to the induction of expressive schemas from large RDF repositories. We describe in detail the implementation of this approach and report on an evaluation that we conducted using several data sets including DBpedia.",
"In this paper we give an overview of the Knowledge Base Population (KBP) track at the 2010 Text Analysis Conference. The main goal of KBP is to promote research in discovering facts about entities and augmenting a knowledge base (KB) with these facts. This is done through two tasks, Entity Linking -- linking names in context to entities in the KB -- and Slot Filling -- adding information about an entity to the KB. A large source collection of newswire and web documents is provided from which systems are to discover information. Attributes (\"slots\") derived from Wikipedia infoboxes are used to create the reference KB. In this paper we provide an overview of the techniques which can serve as a basis for a good KBP system, lay out the remaining challenges by comparison with traditional Information Extraction (IE) and Question Answering (QA) tasks, and provide some suggestions to address these challenges.",
"",
"Recent advances in information extraction have led to huge knowledge bases (KBs), which capture knowledge in a machine-readable format. Inductive Logic Programming (ILP) can be used to mine logical rules from the KB. These rules can help deduce and add missing knowledge to the KB. While ILP is a mature field, mining logical rules from KBs is different in two aspects: First, current rule mining systems are easily overwhelmed by the amount of data (state-of-the art systems cannot even run on today's KBs). Second, ILP usually requires counterexamples. KBs, however, implement the open world assumption (OWA), meaning that absent data cannot be used as counterexamples. In this paper, we develop a rule mining model that is explicitly tailored to support the OWA scenario. It is inspired by association rule mining and introduces a novel measure for confidence. Our extensive experiments show that our approach outperforms state-of-the-art approaches in terms of precision and coverage. Furthermore, our system, AMIE, mines rules orders of magnitude faster than state-of-the-art approaches.",
"DBpedia contains millions of untyped entities, either if we consider the native DBpedia ontology, or Yago plus WordNet. Is it possible to automatically classify those entities? Based on previous work on wikilink invariances, we wondered if wikilinks convey a knowledge rich enough for their classication. In this paper we give three contributions. Concerning the DBpedia link structure, we describe some measurements and notice both problems (e.g. the bias that could be induced by the incomplete ontological coverage of the DBpedia ontology), and potentials existing in current type coverage. Concerning classication, we present two techniques that exploit wikilinks, one based on induction from machine learning techniques, and the other on abduction. Finally, we discuss the limited results of classication, which conrmed our fears expressed in the description of general gures from the measurement. We also suggest some new possible directions to entity classication that could be taken.",
"Over the past few years, massive amounts of world knowledge have been accumulated in publicly available knowledge bases, such as Freebase, NELL, and YAGO. Yet despite their seemingly huge size, these knowledge bases are greatly incomplete. For example, over 70 of people included in Freebase have no known place of birth, and 99 have no known ethnicity. In this paper, we propose a way to leverage existing Web-search-based question-answering technology to fill in the gaps in knowledge bases in a targeted way. In particular, for each entity attribute, we learn the best set of queries to ask, such that the answer snippets returned by the search engine are most likely to contain the correct value for that attribute. For example, if we want to find Frank Zappa's mother, we could ask the query who is the mother of Frank Zappa'. However, this is likely to return The Mothers of Invention', which was the name of his band. Our system learns that it should (in this case) add disambiguating terms, such as Zappa's place of birth, in order to make it more likely that the search results contain snippets mentioning his mother. Our system also learns how many different queries to ask for each attribute, since in some cases, asking too many can hurt accuracy (by introducing false positives). We discuss how to aggregate candidate answers across multiple queries, ultimately returning probabilistic predictions for possible values for each attribute. Finally, we evaluate our system and show that it is able to extract a large number of facts with high confidence."
]
} |
1705.05720 | 2614919369 | Knowledge bases (KBs) have attracted increasing attention due to its great success in various areas, such as Web and mobile search.Existing KBs are restricted to objective factual knowledge, such as city population or fruit shape, whereas,subjective knowledge, such as big city, which is commonly mentioned in Web and mobile queries, has been neglected. Subjective knowledge differs from objective knowledge in that it has no documented or observed ground truth. Instead, the truth relies on people's dominant opinion. Thus, we can use the crowdsourcing technique to get opinion from the crowd. In our work, we propose a system, called crowdsourced subjective knowledge acquisition (CoSKA),for subjective knowledge acquisition powered by crowdsourcing and existing KBs. The acquired knowledge can be used to enrich existing KBs in the subjective dimension which bridges the gap between existing objective knowledge and subjective queries.The main challenge of CoSKA is the conflict between large scale knowledge facts and limited crowdsourcing resource. To address this challenge, in this work, we define knowledge inference rules and then select the seed knowledge judiciously for crowdsourcing to maximize the inference power under the resource constraint. Our experimental results on real knowledge base and crowdsourcing platform verify the effectiveness of CoSKA system. | Recently, the increasing popularity of crowdsourcing brings new trend to leverage the power of the crowd in knowledge acquisition, data integration and many other applications. @cite_27 proposes a hybrid approach that combines information extraction technique with human computation for knowledge acquisition. @cite_3 presents a hybrid-genre workflow for games in crowdsourced knowledge acquisition process. Works @cite_18 @cite_9 @cite_14 present approaches that use the wisdom of crowd to perform taxonomy construction. Crowdsourcing also proved to have good performance in applications such as entity resolution @cite_4 @cite_21 , schema matching @cite_22 , translation @cite_6 and so forth. | {
"cite_N": [
"@cite_18",
"@cite_14",
"@cite_4",
"@cite_22",
"@cite_9",
"@cite_21",
"@cite_3",
"@cite_6",
"@cite_27"
],
"mid": [
"2120396827",
"2243633279",
"2056748234",
"2123885506",
"",
"2106675345",
"",
"2134677057",
"2101349222"
],
"abstract": [
"Taxonomies are a useful and ubiquitous way of organizing information. However, creating organizational hierarchies is difficult because the process requires a global understanding of the objects to be categorized. Usually one is created by an individual or a small group of people working together for hours or even days. Unfortunately, this centralized approach does not work well for the large, quickly changing datasets found on the web. Cascade is an automated workflow that allows crowd workers to spend as little at 20 seconds each while collectively making a taxonomy. We evaluate Cascade and show that on three datasets its quality is 80-90 of that of experts. Cascade has a competitive cost to expert information architects, despite taking six times more human labor. Fortunately, this labor can be parallelized such that Cascade will run in as fast as four minutes instead of hours or days.",
"Recently, taxonomy has attracted much attention. Both automatic construction solutions and human-based computation approaches have been proposed. The automatic methods suffer from the problem of either low precision or low recall and human computation, on the other hand, is not suitable for large scale tasks. Motivated by the shortcomings of both approaches, we present a hybrid framework, which combines the power of machine-based approaches and human computation (the crowd) to construct a more complete and accurate taxonomy. Specifically, our framework consists of two steps: we first construct a complete but noisy taxonomy automatically, then crowd is introducedto adjust the entity positions in the constructed taxonomy. However, the adjustment is challenging as the budget (money) for asking the crowd is often limited. In our work, we formulatethe problem of finding the optimal adjustment as an entityselection optimization (ESO) problem, which is proved to beNP-hard. We then propose an exact algorithm and a moreefficient approximation algorithm with an approximation ratioof 1 2(1-1 e). We conduct extensive experiments on real datasets, the results show that our hybrid approach largely improves the recall of the taxonomy with little impairment for precision.",
"Entity resolution is central to data integration and data cleaning. Algorithmic approaches have been improving in quality, but remain far from perfect. Crowdsourcing platforms offer a more accurate but expensive (and slow) way to bring human insight into the process. Previous work has proposed batching verification tasks for presentation to human workers but even with batching, a human-only approach is infeasible for data sets of even moderate size, due to the large numbers of matches to be tested. Instead, we propose a hybrid human-machine approach in which machines are used to do an initial, coarse pass over all the data, and people are used to verify only the most likely matching pairs. We show that for such a hybrid system, generating the minimum number of verification tasks of a given size is NP-Hard, but we develop a novel two-tiered heuristic approach for creating batched tasks. We describe this method, and present the results of extensive experiments on real data sets using a popular crowdsourcing platform. The experiments show that our hybrid approach achieves both good efficiency and high accuracy compared to machine-only or human-only alternatives.",
"Schema matching is a central challenge for data integration systems. Automated tools are often uncertain about schema matchings they suggest, and this uncertainty is inherent since it arises from the inability of the schema to fully capture the semantics of the represented data. Human common sense can often help. Inspired by the popularity and the success of easily accessible crowdsourcing platforms, we explore the use of crowdsourcing to reduce the uncertainty of schema matching. Since it is typical to ask simple questions on crowdsourcing platforms, we assume that each question, namely Correspondence Correctness Question (CCQ), is to ask the crowd to decide whether a given correspondence should exist in the correct matching. We propose frameworks and efficient algorithms to dynamically manage the CCQs, in order to maximize the uncertainty reduction within a limited budget of questions. We develop two novel approaches, namely \"Single CCQ\" and \"Multiple CCQ\", which adaptively select, publish and manage the questions. We verified the value of our solutions with simulation and real implementation.",
"",
"In this paper, we study a hybrid human-machine approach for solving the problem of Entity Resolution (ER). The goal of ER is to identify all records in a database that refer to the same underlying entity, and are therefore duplicates of each other. Our input is a graph over all the records in a database, where each edge has a probability denoting our prior belief (based on Machine Learning models) that the pair of records represented by the given edge are duplicates. Our objective is to resolve all the duplicates by asking humans to verify the equality of a subset of edges, leveraging the transitivity of the equality relation to infer the remaining edges (e.g. a = c can be inferred given a = b and b = c). We consider the problem of designing optimal strategies for asking questions to humans that minimize the expected number of questions asked. Using our theoretical framework, we analyze several strategies, and show that a strategy, claimed as \"optimal\" for this problem in a recent work, can perform arbitrarily bad in theory. We propose alternate strategies with theoretical guarantees. Using both public datasets as well as the production system at Facebook, we show that our techniques are effective in practice.",
"",
"In this paper, we build a social search engine named Glaucus for location-based queries. They compose a significant portion of mobile searches, thus becoming more popular with the prevalence of mobile devices. However, most of existing social search engines are not designed for location-based queries and thus often produce poor-quality results for such queries. Glaucus is inherently designed to support location-based queries. It collects the check-in information, which pinpoints the places where each user visited, from location-based social networking services such as Foursquare. Then, it calculates the expertise of each user for a query by using our new probabilistic model called the location aspect model . We conducted two types of evaluation to prove the effectiveness of our engine. The results showed that Glaucus selected the users supported by stronger evidence for the required expertise than existing social search engines. In addition, the answers from the experts selected by Glaucus were highly rated by our human judges in terms of answer satisfaction.",
"Automatic information extraction (IE) enables the construction of very large knowledge bases (KBs), with relational facts on millions of entities from text corpora and Web sources. However, such KBs contain errors and they are far from being complete. This motivates the need for exploiting human intelligence and knowledge using crowd-based human computing (HC) for assessing the validity of facts and for gathering additional knowledge. This paper presents a novel system architecture, called Higgins, which shows how to effectively integrate an IE engine and a HC engine. Higgins generates game questions where players choose or fill in missing relations for subject-relation-object triples. For generating multiple-choice answer candidates, we have constructed a large dictionary of entity names and relational phrases, and have developed specifically designed statistical language models for phrase relatedness. To this end, we combine semantic resources like WordNet, ConceptNet, and others with statistics derived from a largeWeb corpus. We demonstrate the effectiveness of Higgins for knowledge acquisition by crowdsourced gathering of relationships between characters in narrative descriptions of movies and books."
]
} |
1705.05363 | 2614839826 | In many real-world scenarios, rewards extrinsic to the agent are extremely sparse, or absent altogether. In such cases, curiosity can serve as an intrinsic reward signal to enable the agent to explore its environment and learn skills that might be useful later in its life. We formulate curiosity as the error in an agent's ability to predict the consequence of its own actions in a visual feature space learned by a self-supervised inverse dynamics model. Our formulation scales to high-dimensional continuous state spaces like images, bypasses the difficulties of directly predicting pixels, and, critically, ignores the aspects of the environment that cannot affect the agent. The proposed approach is evaluated in two environments: VizDoom and Super Mario Bros. Three broad settings are investigated: 1) sparse extrinsic reward, where curiosity allows for far fewer interactions with the environment to reach the goal; 2) exploration with no extrinsic reward, where curiosity pushes the agent to explore more efficiently; and 3) generalization to unseen scenarios (e.g. new levels of the same game) where the knowledge gained from earlier experience helps the agent explore new places much faster than starting from scratch. Demo video and code available at this https URL | Concurrent work: A number of interesting related papers have appeared on Arxiv while the present work was in submission. sukhbaatar2017intrinsic generates supervision for pre-training via asymmetric self-play between two agents to improve data efficiency during fine-tuning. Several methods propose improving data efficiency of RL algorithms using self-supervised prediction based auxiliary tasks @cite_35 @cite_5 . fu2017ex2 learn discriminative models, and gregor2017variational use empowerment based measure to tackle exploration in sparse reward setups. | {
"cite_N": [
"@cite_35",
"@cite_5"
],
"mid": [
"2551887912",
"2563830277"
],
"abstract": [
"Deep reinforcement learning agents have achieved state-of-the-art results by directly maximising cumulative reward. However, environments contain a much wider variety of possible training signals. In this paper, we introduce an agent that also maximises many other pseudo-reward functions simultaneously by reinforcement learning. All of these tasks share a common representation that, like unsupervised learning, continues to develop in the absence of extrinsic rewards. We also introduce a novel mechanism for focusing this representation upon extrinsic rewards, so that learning can rapidly adapt to the most relevant aspects of the actual task. Our agent significantly outperforms the previous state-of-the-art on Atari, averaging 880 expert human performance, and a challenging suite of first-person, three-dimensional tasks leading to a mean speedup in learning of 10 @math and averaging 87 expert human performance on Labyrinth.",
"Reinforcement learning optimizes policies for expected cumulative reward. Need the supervision be so narrow? Reward is delayed and sparse for many tasks, making it a difficult and impoverished signal for end-to-end optimization. To augment reward, we consider a range of self-supervised tasks that incorporate states, actions, and successors to provide auxiliary losses. These losses offer ubiquitous and instantaneous supervision for representation learning even in the absence of reward. While current results show that learning from reward alone is feasible, pure reinforcement learning methods are constrained by computational and data efficiency issues that can be remedied by auxiliary losses. Self-supervised pre-training and joint optimization improve the data efficiency and policy returns of end-to-end reinforcement learning."
]
} |
1705.05316 | 2614581251 | Local consistencies stronger than arc consistency have received a lot of attention since the early days of CSP research. because of the strong pruning they can achieve. However, they have not been widely adopted by CSP solvers. This is because applying such consistencies can sometimes result in considerably smaller search tree sizes and therefore in important speed-ups, but in other cases the search space reduction may be small, causing severe run time penalties. Taking advantage of recent advances in parallelization, we propose a novel approach for the application of strong local consistencies (SLCs) that can improve their performance by largely preserving the speed-ups they offer in cases where they are successful, and eliminating the run time penalties in cases where they are unsuccessful. This approach is presented in the form of two search algorithms. Both algorithms consist of a master search process, which is a typical CSP solver, and a number of slave processes, with each one implementing a SLC method. The first algorithm runs the different SLCs synchronously at each node of the search tree explored in the master process, while the second one can run them asynchronously at different nodes of the search tree. Experimental results demonstrate the benefits of the proposed method. | There is a quite extensive body of work on parallel constraint solving which aims at exploiting the increasing number of available processors to speed up computation. A review can be found in the paper by @cite_29 . Such works are relevant to our framework since they also exploit multiple processors, but at the same time they are quite different. Parallel CSP (and SAT) solving has mainly focused on search space splitting (i.e. allocating different branches of the search tree to different processors), e.g. the works by @cite_22 @cite_11 @cite_5 @cite_9 @cite_8 @cite_21 , and solver portfolios, e.g. the works by @cite_2 @cite_24 @cite_25 @cite_26 @cite_27 , and to a lesser extent, on the parallelization of propagation, e.g. the works by @cite_19 @cite_3 @cite_16 . | {
"cite_N": [
"@cite_26",
"@cite_22",
"@cite_8",
"@cite_29",
"@cite_9",
"@cite_21",
"@cite_3",
"@cite_24",
"@cite_19",
"@cite_27",
"@cite_2",
"@cite_5",
"@cite_16",
"@cite_25",
"@cite_11"
],
"mid": [
"153192071",
"",
"158822462",
"2730963839",
"",
"2281349988",
"110346337",
"",
"2067633747",
"2058491973",
"2103406309",
"2136680084",
"1765933479",
"37608273",
"2107243813"
],
"abstract": [
"Parallelization offers the opportunity to accelerate search on constraint satisfaction problems. To parallelize a sequential solver under a popular message passing protocol, the new paradigm described here combines portfolio-based methods and search space splitting. To split effectively and to balance processor workload, this paradigm adaptively exploits knowledge acquired during search and allocates additional resources to the most difficult parts of a problem. Extensive experiments in a parallel environment show that this paradigm significantly improves the performance of an underlying sequential solver, outperforms more naive approaches to parallelization, and solves many difficult problems left open after recent solver competitions.",
"",
"The computing industry is currently facing a major architectural shift. Extra computing power is not coming anymore from higher processor frequencies, but from a growing number of computing cores and processors. For AI, and constraint solving in particular, this raises the question of how to scale current solving techniques to massively parallel architectures. While prior work focusses mostly on small scale parallel constraint solving, we conduct the first study on scalability of constraint solving on 100 processors and beyond in this paper. We propose techniques that are simple to apply and show empirically that they scale surprisingly well. These techniques establish a performance baseline for parallel constraint solving technologies against which more sophisticated parallel algorithms need to compete in the future.",
"With the ubiquity of multicore computing, and the likely expansion of it, it seems irresponsible for constraints researchers to ignore the implications of it. Therefore, the authors have recently begun investigating the literature in constraints on exploitation of parallel systems for constraint solving. We have been compiling an incomplete, biased, and ill-written review of this literature. While accepting these faults, we nevertheless hope that it may provide some useful pointers to others wishing to follow a similar path to us: that is a path from complete to only partial ignorance.",
"",
"We propose the Embarrassingly Parallel Search, a simple and efficient method for solving constraint programming problems in parallel. We split the initial problem into a huge number of independent subproblems and solve them with available workers, for instance cores of machines. The decomposition into subproblems is computed by selecting a subset of variables and by enumerating the combinations of values of these variables that are not detected inconsistent by the propagation mechanism of a CP Solver. The experiments on satisfaction problems and optimization problems suggest that generating between thirty and one hundred subproblems per worker leads to a good scalability. We show that our method is quite competitive with the work stealing approach and able to solve some classical problems at the maximum capacity of the multi-core machines. Thanks to it, a user can parallelize the resolution of its problem without modifying the solver or writing any parallel source code and can easily replay the resolution of a problem.",
"",
"",
"Abstract Constraint satisfaction networks have been shown to be a very useful tool for knowledge representation in Artificial Intelligence applications. These networks often utilize local constraint propagation techniques to achieve local consistency (consistent labeling in vision). Such methods have been used extensively in the context of image understanding and interpretation, as well as planning, natural language analysis and truth maintenance systems. In this paper we study the parallel complexity of discrete relaxation, one of the most commonly used constraint propagation techniques. Since the constraint propagation procedures such as discrete relaxation appear to operate locally, it has been previously believed that the relaxation approach for achieving local consistency has a natural parallel solution. Our analysis suggests that a parallel solution is unlikely to improve the known sequential solutions by much. Specifically, we prove that the problem solved by discrete relaxation (arc consistency) is log-space complete for P (the class of polynomial-time deterministic sequential algorithms). Intuitively, this implies that discrete relaxation is inherently sequential and it is unlikely that we can solve the polynomial-time version of the consistent labeling problem in logarithmic time by using only a polynomial number of processors. Some practical implications of our result are discussed. We also provide a two-way transformation between AND OR graphs, propositional Horn satisfiability and local consistency in constraint networks that allows us to develop optimal linear-time algorithms for local consistency in constraint networks.",
"Portfolio based approaches to constraint solving aim at exploiting the variability in performance displayed by different solvers or different parameter settings of a single solver. Such approaches have been quite successful in both a sequential and a parallel processing mode. Given the increasingly larger number of available processors for parallel processing, an important challenge when designing portfolios is to identify solver parameters that offer diversity in the exploration of the search space and to generate different solver configurations by automatically tuning these parameters. In this paper we propose, for the first time, a way to build porfolios for parallel solving by parameter zing the local consistency property applied during search. To achieve this we exploit heuristics for adaptive propagation proposed in stergiou08. We show how this approach can result in the easy automatic generation of portfolios that display large performance variability. We make an experimental comparison against a standard sequential solver as well as portfolio based methods that use randomization of the variable ordering heuristic as the source of diversity. Results demonstrate that our method constantly outperforms the sequential solver and in most cases it is more efficient than the other portfolio approaches.",
"In this paper, ManySAT a new portfolio-based parallel SAT solver is thoroughly described. The design of ManySAT benefits from the main weaknesses of modern SAT solvers: their sensitivity to parameter tuning and their lack of robustness. ManySAT uses a portfolio of complementary sequential algorithms obtained through careful variations of the standard DPLL algorithm. Additionally, each sequential algorithm shares clauses to improve the overall performance of the whole system. This contrasts with most of the parallel SAT solvers generally designed using the divide-and-conquer paradigm. Experiments on many industrial SAT instances, and the first rank obtained by ManySAT in the parallel track of the 2008 SAT-Race clearly show the potential of our design philosophy.",
"The availability of commodity multicore and multiprocessor machines and the inherent parallelism in constraint programming search offer significant opportunities for constraint programming. These opportunities also present a fundamental challenge: how to exploit parallelism transparently to speed up constraint programs. This paper shows how to parallelize constraint programs transparently without changes to the sequential code. The main technical idea consists of automatically lifting a sequential exploration strategy into its parallel counterpart, allowing workers to share and steal subproblems. Experimental results show that the parallel implementation may produce significant speedups on multicore machines.",
"Program parallelization becomes increasingly important when new multi-core architectures provide ways to improve performance. One of the greatest challenges of this development lies in programming parallel applications. Declarative languages, such as constraint programming, can make the transition to parallelism easier by hiding the parallelization details in a framework. Automatic parallelization in constraint programming has mostly focused on parallel search. While search and consistency are intrinsically linked, the consistency part of the solving process is often more time-consuming. We have previously looked at parallel consistency and found it to be quite promising. In this paper we investigate how to combine parallel search with parallel consistency. We evaluate which problems are suitable and which are not. Our results show that parallelizing the entire solving process in constraint programming is a major challenge as parallel search and parallel consistency typically suit different types of problems. (Less)",
"Managing learnt clauses among a parallel, memory shared, SAT solver is a crucial but difficult task. Based on some statistical experiments made on learnt clauses, we propose a simple parallel version of Glucose that uses a lazy policy to exchange clauses between cores. This policy does not send a clause when it is learnt, but later, when it has a chance to be useful locally. We also propose a strategy for clauses importation that put them in ”probation” before a potential entry in the search, thus limiting the negative impact of high importation rates, both in terms of noise and decreasing propagation speed.",
"We present a framework for the parallelization of depth-first combinatorial search algorithms on a network of computers. Our architecture is intended for a distributed setting and uses a work stealing strategy coupled with a small number of primitives for the processors (which we call workers) to obtain new work and to communicate to other workers. These primitives are a minimal imposition and integrate easily with constraint programming systems. The main contribution is an adaptive architecture, which allows workers to incrementally join and leave and has good scaling properties as the number of workers increases. Our empirical results illustrate that near-linear speedup for backtrack search is achieved for up to 61 workers. It suggests that near-linear speedup is possible with even more workers. The experiments also demonstrate where departures from linearity can occur for small problems, and also for problems where the parallelism can itself affect the search as in branch and bound."
]
} |
1705.05316 | 2614581251 | Local consistencies stronger than arc consistency have received a lot of attention since the early days of CSP research. because of the strong pruning they can achieve. However, they have not been widely adopted by CSP solvers. This is because applying such consistencies can sometimes result in considerably smaller search tree sizes and therefore in important speed-ups, but in other cases the search space reduction may be small, causing severe run time penalties. Taking advantage of recent advances in parallelization, we propose a novel approach for the application of strong local consistencies (SLCs) that can improve their performance by largely preserving the speed-ups they offer in cases where they are successful, and eliminating the run time penalties in cases where they are unsuccessful. This approach is presented in the form of two search algorithms. Both algorithms consist of a master search process, which is a typical CSP solver, and a number of slave processes, with each one implementing a SLC method. The first algorithm runs the different SLCs synchronously at each node of the search tree explored in the master process, while the second one can run them asynchronously at different nodes of the search tree. Experimental results demonstrate the benefits of the proposed method. | Regarding the latter direction, which is closer to our work, parallelizing constraint propagation algorithms is a challenging task since most such algorithms are sequential by nature, as demonstrated by @cite_19 . Hence, this approach has not been explored as much as the other ones, and it is quite different to our work where each LC algorithm runs on a single processor. Another common perception that has resulted in limited research on constraint propagation parallelization is that the scalability of this approach is limited by Amdahl's law: "if propagation consumes 80$ Existing works on parallel constraint propagation have focused on AC and have either been purely theoretical, or any experiments that were conducted, e.g. by @cite_3 and @cite_15 either failed to show significant speed-ups or were limited to very few processors. @cite_16 consider the parallelization of a modern CP solvers' constraint propagation engine and shows that problems with a large number of (expensive to propagate) global constraints can benefit from parallelization of the propagation mechanism. Since this approach is orthogonal to ours, their combination is an interesting avenue for research. | {
"cite_N": [
"@cite_19",
"@cite_15",
"@cite_16",
"@cite_3"
],
"mid": [
"2067633747",
"2086991642",
"1765933479",
"110346337"
],
"abstract": [
"Abstract Constraint satisfaction networks have been shown to be a very useful tool for knowledge representation in Artificial Intelligence applications. These networks often utilize local constraint propagation techniques to achieve local consistency (consistent labeling in vision). Such methods have been used extensively in the context of image understanding and interpretation, as well as planning, natural language analysis and truth maintenance systems. In this paper we study the parallel complexity of discrete relaxation, one of the most commonly used constraint propagation techniques. Since the constraint propagation procedures such as discrete relaxation appear to operate locally, it has been previously believed that the relaxation approach for achieving local consistency has a natural parallel solution. Our analysis suggests that a parallel solution is unlikely to improve the known sequential solutions by much. Specifically, we prove that the problem solved by discrete relaxation (arc consistency) is log-space complete for P (the class of polynomial-time deterministic sequential algorithms). Intuitively, this implies that discrete relaxation is inherently sequential and it is unlikely that we can solve the polynomial-time version of the consistent labeling problem in logarithmic time by using only a polynomial number of processors. Some practical implications of our result are discussed. We also provide a two-way transformation between AND OR graphs, propositional Horn satisfiability and local consistency in constraint networks that allows us to develop optimal linear-time algorithms for local consistency in constraint networks.",
"Consistency techniques are an efficient way of tackling constraint satisfaction problems (CSP). In particular, various are-consistency algorithms have been designed such as the rime optimal AC-4 sequential algorithm of Mohr and Henderson (1986). In this paper, we present a new distributed are-consistency algorithm, called DisAC-4. DisAC-4 is based on AC-4, and is a coarse-grained parallel algorithm designed for distributed memory computers using message passing communication. Termination and correctness of the algorithm are proven. Theoretical complexities and experimental results are given. Both show linear speedup with respect to the number of processors. The strong point of DisAC-4 is its suitability to be implemented on very common hardware infrastructures like networks of workstations and or PCs as well as on intensive computing parallel mainframes. (C) 1998 Elsevier Science B.V.",
"Program parallelization becomes increasingly important when new multi-core architectures provide ways to improve performance. One of the greatest challenges of this development lies in programming parallel applications. Declarative languages, such as constraint programming, can make the transition to parallelism easier by hiding the parallelization details in a framework. Automatic parallelization in constraint programming has mostly focused on parallel search. While search and consistency are intrinsically linked, the consistency part of the solving process is often more time-consuming. We have previously looked at parallel consistency and found it to be quite promising. In this paper we investigate how to combine parallel search with parallel consistency. We evaluate which problems are suitable and which are not. Our results show that parallelizing the entire solving process in constraint programming is a major challenge as parallel search and parallel consistency typically suit different types of problems. (Less)",
""
]
} |
1705.05483 | 2951183746 | In recent years, text recognition has achieved remarkable success in recognizing scanned document text. However, word recognition in natural images is still an open problem, which generally requires time consuming post-processing steps. We present a novel architecture for individual word detection in scene images based on semantic segmentation. Our contributions are twofold: the concept of WordFence, which detects border areas surrounding each individual word and a novel pixelwise weighted softmax loss function which penalizes background and emphasizes small text regions. WordFence ensures that each word is detected individually, and the new loss function provides a strong training signal to both text and word border localization. The proposed technique avoids intensive post-processing, producing an end-to-end word detection system. We achieve superior localization recall on common benchmark datasets - 92 recall on ICDAR11 and ICDAR13 and 63 recall on SVT. Furthermore, our end-to-end word recognition system achieves state-of-the-art 86 F-Score on ICDAR13. | A major limitation of CNNs is that networks have trouble taking different scales of images into account. Networks generally use max-pooling layers to reduce the search space for training - this operation reduces resolution and loses spatial information between different features. Yu and Koltun @cite_16 argued that max-pooling does not maintain global scale information and propose dilated convolutions to increase the effective receptive field of convolutional operations. Other works tackled the scale problem with methods such as fully convolutional networks (FCNs) @cite_29 or with atrous convolutions @cite_20 @cite_1 . Another challenge addressed by CNNs is semantic segmentation - where each pixel in the image has to be matched to a specific label. Semantic segmentation has recently been enhanced by dilated convolutions @cite_16 , FCNs @cite_29 and probabilistic graphical models @cite_33 . | {
"cite_N": [
"@cite_33",
"@cite_29",
"@cite_1",
"@cite_16",
"@cite_20"
],
"mid": [
"1923697677",
"2952632681",
"",
"2286929393",
"1487583988"
],
"abstract": [
"Deep Convolutional Neural Networks (DCNNs) have recently shown state of the art performance in high level vision tasks, such as image classification and object detection. This work brings together methods from DCNNs and probabilistic graphical models for addressing the task of pixel-level classification (also called \"semantic image segmentation\"). We show that responses at the final layer of DCNNs are not sufficiently localized for accurate object segmentation. This is due to the very invariance properties that make DCNNs good for high level tasks. We overcome this poor localization property of deep networks by combining the responses at the final DCNN layer with a fully connected Conditional Random Field (CRF). Qualitatively, our \"DeepLab\" system is able to localize segment boundaries at a level of accuracy which is beyond previous methods. Quantitatively, our method sets the new state-of-art at the PASCAL VOC-2012 semantic image segmentation task, reaching 71.6 IOU accuracy in the test set. We show how these results can be obtained efficiently: Careful network re-purposing and a novel application of the 'hole' algorithm from the wavelet community allow dense computation of neural net responses at 8 frames per second on a modern GPU.",
"Convolutional networks are powerful visual models that yield hierarchies of features. We show that convolutional networks by themselves, trained end-to-end, pixels-to-pixels, exceed the state-of-the-art in semantic segmentation. Our key insight is to build \"fully convolutional\" networks that take input of arbitrary size and produce correspondingly-sized output with efficient inference and learning. We define and detail the space of fully convolutional networks, explain their application to spatially dense prediction tasks, and draw connections to prior models. We adapt contemporary classification networks (AlexNet, the VGG net, and GoogLeNet) into fully convolutional networks and transfer their learned representations by fine-tuning to the segmentation task. We then define a novel architecture that combines semantic information from a deep, coarse layer with appearance information from a shallow, fine layer to produce accurate and detailed segmentations. Our fully convolutional network achieves state-of-the-art segmentation of PASCAL VOC (20 relative improvement to 62.2 mean IU on 2012), NYUDv2, and SIFT Flow, while inference takes one third of a second for a typical image.",
"",
"State-of-the-art models for semantic segmentation are based on adaptations of convolutional networks that had originally been designed for image classification. However, dense prediction and image classification are structurally different. In this work, we develop a new convolutional network module that is specifically designed for dense prediction. The presented module uses dilated convolutions to systematically aggregate multi-scale contextual information without losing resolution. The architecture is based on the fact that dilated convolutions support exponential expansion of the receptive field without loss of resolution or coverage. We show that the presented context module increases the accuracy of state-of-the-art semantic segmentation systems. In addition, we examine the adaptation of image classification networks to dense prediction and show that simplifying the adapted network can increase accuracy.",
"We present an integrated framework for using Convolutional Networks for classification, localization and detection. We show how a multiscale and sliding window approach can be efficiently implemented within a ConvNet. We also introduce a novel deep learning approach to localization by learning to predict object boundaries. Bounding boxes are then accumulated rather than suppressed in order to increase detection confidence. We show that different tasks can be learned simultaneously using a single shared network. This integrated framework is the winner of the localization task of the ImageNet Large Scale Visual Recognition Challenge 2013 (ILSVRC2013) and obtained very competitive results for the detection and classifications tasks. In post-competition work, we establish a new state of the art for the detection task. Finally, we release a feature extractor from our best model called OverFeat."
]
} |
1705.05483 | 2951183746 | In recent years, text recognition has achieved remarkable success in recognizing scanned document text. However, word recognition in natural images is still an open problem, which generally requires time consuming post-processing steps. We present a novel architecture for individual word detection in scene images based on semantic segmentation. Our contributions are twofold: the concept of WordFence, which detects border areas surrounding each individual word and a novel pixelwise weighted softmax loss function which penalizes background and emphasizes small text regions. WordFence ensures that each word is detected individually, and the new loss function provides a strong training signal to both text and word border localization. The proposed technique avoids intensive post-processing, producing an end-to-end word detection system. We achieve superior localization recall on common benchmark datasets - 92 recall on ICDAR11 and ICDAR13 and 63 recall on SVT. Furthermore, our end-to-end word recognition system achieves state-of-the-art 86 F-Score on ICDAR13. | In the aforementioned scenario, text localization is considered to be the key task, since a well-cropped proposal can be fed to a word recognition system @cite_15 . Before CNNs, popular methods for text localization utilized computer vision techniques with hand-crafted feature descriptors. More recent works have used CNN features. However, all of these approaches have a limitation of feature driven engineering - there are simply too many edge cases to account for. The detectors generate a large amount of non-text false positives, requiring additional filtering techniques. Often, a number of post-processing steps is needed to reach a good performance. | {
"cite_N": [
"@cite_15"
],
"mid": [
"1491389626"
],
"abstract": [
"In this work we present a framework for the recognition of natural scene text. Our framework does not require any human-labelled data, and performs word recognition on the whole image holistically, departing from the character based recognition systems of the past. The deep neural network models at the centre of this framework are trained solely on data produced by a synthetic text generation engine -- synthetic data that is highly realistic and sufficient to replace real data, giving us infinite amounts of training data. This excess of data exposes new possibilities for word recognition models, and here we consider three models, each one \"reading\" words in a different way: via 90k-way dictionary encoding, character sequence encoding, and bag-of-N-grams encoding. In the scenarios of language based and completely unconstrained text recognition we greatly improve upon state-of-the-art performance on standard datasets, using our fast, simple machinery and requiring zero data-acquisition costs."
]
} |
1705.05483 | 2951183746 | In recent years, text recognition has achieved remarkable success in recognizing scanned document text. However, word recognition in natural images is still an open problem, which generally requires time consuming post-processing steps. We present a novel architecture for individual word detection in scene images based on semantic segmentation. Our contributions are twofold: the concept of WordFence, which detects border areas surrounding each individual word and a novel pixelwise weighted softmax loss function which penalizes background and emphasizes small text regions. WordFence ensures that each word is detected individually, and the new loss function provides a strong training signal to both text and word border localization. The proposed technique avoids intensive post-processing, producing an end-to-end word detection system. We achieve superior localization recall on common benchmark datasets - 92 recall on ICDAR11 and ICDAR13 and 63 recall on SVT. Furthermore, our end-to-end word recognition system achieves state-of-the-art 86 F-Score on ICDAR13. | With the prominence of deep learning, CNN based regression of candidate bounding boxes has started becoming utilized for filtering false positive candidates. Bounding box detection has been proposed in the context of object detection by works such as You Only Look Once (YOLO) @cite_3 , Faster-RCNN (F-RCNN) @cite_5 and SSD: Single Shot MultiBox Detector @cite_13 . Advances in semantic segmentation @cite_34 @cite_17 have allowed dense prediction to provide input to bounding box regressors. Building on successful implementations of CNNs for semantic segmentation using FCNs for dense prediction @cite_29 , several researchers have introduced object localization via FCNs @cite_11 . | {
"cite_N": [
"@cite_11",
"@cite_29",
"@cite_3",
"@cite_5",
"@cite_34",
"@cite_13",
"@cite_17"
],
"mid": [
"2950800384",
"2952632681",
"",
"",
"2952365771",
"2193145675",
"2333563142"
],
"abstract": [
"We present region-based, fully convolutional networks for accurate and efficient object detection. In contrast to previous region-based detectors such as Fast Faster R-CNN that apply a costly per-region subnetwork hundreds of times, our region-based detector is fully convolutional with almost all computation shared on the entire image. To achieve this goal, we propose position-sensitive score maps to address a dilemma between translation-invariance in image classification and translation-variance in object detection. Our method can thus naturally adopt fully convolutional image classifier backbones, such as the latest Residual Networks (ResNets), for object detection. We show competitive results on the PASCAL VOC datasets (e.g., 83.6 mAP on the 2007 set) with the 101-layer ResNet. Meanwhile, our result is achieved at a test-time speed of 170ms per image, 2.5-20x faster than the Faster R-CNN counterpart. Code is made publicly available at: this https URL",
"Convolutional networks are powerful visual models that yield hierarchies of features. We show that convolutional networks by themselves, trained end-to-end, pixels-to-pixels, exceed the state-of-the-art in semantic segmentation. Our key insight is to build \"fully convolutional\" networks that take input of arbitrary size and produce correspondingly-sized output with efficient inference and learning. We define and detail the space of fully convolutional networks, explain their application to spatially dense prediction tasks, and draw connections to prior models. We adapt contemporary classification networks (AlexNet, the VGG net, and GoogLeNet) into fully convolutional networks and transfer their learned representations by fine-tuning to the segmentation task. We then define a novel architecture that combines semantic information from a deep, coarse layer with appearance information from a shallow, fine layer to produce accurate and detailed segmentations. Our fully convolutional network achieves state-of-the-art segmentation of PASCAL VOC (20 relative improvement to 62.2 mean IU on 2012), NYUDv2, and SIFT Flow, while inference takes one third of a second for a typical image.",
"",
"",
"In this paper, we propose a novel approach for text detec- tion in natural images. Both local and global cues are taken into account for localizing text lines in a coarse-to-fine pro- cedure. First, a Fully Convolutional Network (FCN) model is trained to predict the salient map of text regions in a holistic manner. Then, text line hypotheses are estimated by combining the salient map and character components. Fi- nally, another FCN classifier is used to predict the centroid of each character, in order to remove the false hypotheses. The framework is general for handling text in multiple ori- entations, languages and fonts. The proposed method con- sistently achieves the state-of-the-art performance on three text detection benchmarks: MSRA-TD500, ICDAR2015 and ICDAR2013.",
"We present a method for detecting objects in images using a single deep neural network. Our approach, named SSD, discretizes the output space of bounding boxes into a set of default boxes over different aspect ratios and scales per feature map location. At prediction time, the network generates scores for the presence of each object category in each default box and produces adjustments to the box to better match the object shape. Additionally, the network combines predictions from multiple feature maps with different resolutions to naturally handle objects of various sizes. SSD is simple relative to methods that require object proposals because it completely eliminates proposal generation and subsequent pixel or feature resampling stages and encapsulates all computation in a single network. This makes SSD easy to train and straightforward to integrate into systems that require a detection component. Experimental results on the PASCAL VOC, COCO, and ILSVRC datasets confirm that SSD has competitive accuracy to methods that utilize an additional object proposal step and is much faster, while providing a unified framework for both training and inference. For (300 300 ) input, SSD achieves 74.3 mAP on VOC2007 test at 59 FPS on a Nvidia Titan X and for (512 512 ) input, SSD achieves 76.9 mAP, outperforming a comparable state of the art Faster R-CNN model. Compared to other single stage methods, SSD has much better accuracy even with a smaller input image size. Code is available at https: github.com weiliu89 caffe tree ssd.",
"We introduce a new top-down pipeline for scene text detection. We propose a novel Cascaded Convolutional Text Network (CCTN) that joints two customized convolutional networks for coarse-to-fine text localization. The CCTN fast detects text regions roughly from a low-resolution image, and then accurately localizes text lines from each enlarged region. We cast previous character based detection into direct text region estimation, avoiding multiple bottom- up post-processing steps. It exhibits surprising robustness and discriminative power by considering whole text region as detection object which provides strong semantic information. We customize convolutional network by develop- ing rectangle convolutions and multiple in-network fusions. This enables it to handle multi-shape and multi-scale text efficiently. Furthermore, the CCTN is computationally efficient by sharing convolutional computations, and high-level property allows it to be invariant to various languages and multiple orientations. It achieves 0.84 and 0.86 F-measures on the ICDAR 2011 and ICDAR 2013, delivering substantial improvements over state-of-the-art results [23, 1]."
]
} |
1705.05483 | 2951183746 | In recent years, text recognition has achieved remarkable success in recognizing scanned document text. However, word recognition in natural images is still an open problem, which generally requires time consuming post-processing steps. We present a novel architecture for individual word detection in scene images based on semantic segmentation. Our contributions are twofold: the concept of WordFence, which detects border areas surrounding each individual word and a novel pixelwise weighted softmax loss function which penalizes background and emphasizes small text regions. WordFence ensures that each word is detected individually, and the new loss function provides a strong training signal to both text and word border localization. The proposed technique avoids intensive post-processing, producing an end-to-end word detection system. We achieve superior localization recall on common benchmark datasets - 92 recall on ICDAR11 and ICDAR13 and 63 recall on SVT. Furthermore, our end-to-end word recognition system achieves state-of-the-art 86 F-Score on ICDAR13. | Early work by Zhang al @cite_34 used a semantic segmentation model to extract text proposals and refine them by applying hand-crafted heuristics. He al @cite_17 improved on previous approaches by introducing a cascade of networks. Gupta al @cite_9 adapted YOLO's approach @cite_3 for text detection and introduced SynthText - a new synthetic text dataset for training. Analogously, F-RCNN @cite_5 was adapted for text recognition by Zhong al @cite_28 and Tian al @cite_7 . The former integrated the F-RCNN framework into a more powerful model. However, a large number of proposals needed to be filtered with a time consuming process. Tian al @cite_7 fused F-RCNN with a recurrent neural network (RNN), allowing the RNN to consider the proposals as a sequence. | {
"cite_N": [
"@cite_7",
"@cite_28",
"@cite_9",
"@cite_3",
"@cite_5",
"@cite_34",
"@cite_17"
],
"mid": [
"2519818067",
"2395360388",
"2952302849",
"",
"",
"2952365771",
"2333563142"
],
"abstract": [
"We propose a novel Connectionist Text Proposal Network (CTPN) that accurately localizes text lines in natural image. The CTPN detects a text line in a sequence of fine-scale text proposals directly in convolutional feature maps. We develop a vertical anchor mechanism that jointly predicts location and text non-text score of each fixed-width proposal, considerably improving localization accuracy. The sequential proposals are naturally connected by a recurrent neural network, which is seamlessly incorporated into the convolutional network, resulting in an end-to-end trainable model. This allows the CTPN to explore rich context information of image, making it powerful to detect extremely ambiguous text. The CTPN works reliably on multi-scale and multi-language text without further post-processing, departing from previous bottom-up methods requiring multi-step post filtering. It achieves 0.88 and 0.61 F-measure on the ICDAR 2013 and 2015 benchmarks, surpassing recent results [8, 35] by a large margin. The CTPN is computationally efficient with 0.14 s image, by using the very deep VGG16 model [27]. Online demo is available: http: textdet.com .",
"In this paper, we develop a novel unified framework called DeepText for text region proposal generation and text detection in natural images via a fully convolutional neural network (CNN). First, we propose the inception region proposal network (Inception-RPN) and design a set of text characteristic prior bounding boxes to achieve high word recall with only hundred level candidate proposals. Next, we present a powerful textdetection network that embeds ambiguous text category (ATC) information and multilevel region-of-interest pooling (MLRP) for text and non-text classification and accurate localization. Finally, we apply an iterative bounding box voting scheme to pursue high recall in a complementary manner and introduce a filtering algorithm to retain the most suitable bounding box, while removing redundant inner and outer boxes for each text instance. Our approach achieves an F-measure of 0.83 and 0.85 on the ICDAR 2011 and 2013 robust text detection benchmarks, outperforming previous state-of-the-art results.",
"In this paper we introduce a new method for text detection in natural images. The method comprises two contributions: First, a fast and scalable engine to generate synthetic images of text in clutter. This engine overlays synthetic text to existing background images in a natural way, accounting for the local 3D scene geometry. Second, we use the synthetic images to train a Fully-Convolutional Regression Network (FCRN) which efficiently performs text detection and bounding-box regression at all locations and multiple scales in an image. We discuss the relation of FCRN to the recently-introduced YOLO detector, as well as other end-to-end object detection systems based on deep learning. The resulting detection network significantly out performs current methods for text detection in natural images, achieving an F-measure of 84.2 on the standard ICDAR 2013 benchmark. Furthermore, it can process 15 images per second on a GPU.",
"",
"",
"In this paper, we propose a novel approach for text detec- tion in natural images. Both local and global cues are taken into account for localizing text lines in a coarse-to-fine pro- cedure. First, a Fully Convolutional Network (FCN) model is trained to predict the salient map of text regions in a holistic manner. Then, text line hypotheses are estimated by combining the salient map and character components. Fi- nally, another FCN classifier is used to predict the centroid of each character, in order to remove the false hypotheses. The framework is general for handling text in multiple ori- entations, languages and fonts. The proposed method con- sistently achieves the state-of-the-art performance on three text detection benchmarks: MSRA-TD500, ICDAR2015 and ICDAR2013.",
"We introduce a new top-down pipeline for scene text detection. We propose a novel Cascaded Convolutional Text Network (CCTN) that joints two customized convolutional networks for coarse-to-fine text localization. The CCTN fast detects text regions roughly from a low-resolution image, and then accurately localizes text lines from each enlarged region. We cast previous character based detection into direct text region estimation, avoiding multiple bottom- up post-processing steps. It exhibits surprising robustness and discriminative power by considering whole text region as detection object which provides strong semantic information. We customize convolutional network by develop- ing rectangle convolutions and multiple in-network fusions. This enables it to handle multi-shape and multi-scale text efficiently. Furthermore, the CCTN is computationally efficient by sharing convolutional computations, and high-level property allows it to be invariant to various languages and multiple orientations. It achieves 0.84 and 0.86 F-measures on the ICDAR 2011 and ICDAR 2013, delivering substantial improvements over state-of-the-art results [23, 1]."
]
} |
1705.05326 | 2615123031 | We develop the theory and practice of an approach to modelling and probabilistic inference in causal networks that is suitable when application-specific or analysis-specific constraints should inform such inference or when little or no data for the learning of causal network structure or probability values at nodes are available. Constrained Bayesian Networks generalize a Bayesian Network such that probabilities can be symbolic, arithmetic expressions and where the meaning of the network is constrained by finitely many formulas from the theory of the reals. A formal semantics for constrained Bayesian Networks over first-order logic of the reals is given, which enables non-linear and non-convex optimisation algorithms that rely on decision procedures for this logic, and supports the composition of several constrained Bayesian Networks. A non-trivial case study in arms control, where few or no data are available to assess the effectiveness of an arms inspection process, evaluates our approach. An open-access prototype implementation of these foundations and their algorithms uses the SMT solver Z3 as decision procedure, leverages an open-source package for Bayesian inference to symbolic computation, and is evaluated experimentally. | -- see e.g. @cite_37 -- refer to the theory and practice of associating a convex set of probability measures with directed, acyclic graphs. Credal networks are also referred to as the Theory of Imprecise Probabilities @cite_1 or as the Quasi-Bayesian Theory @cite_29 . | {
"cite_N": [
"@cite_37",
"@cite_29",
"@cite_1"
],
"mid": [
"",
"1964504235",
"1968293352"
],
"abstract": [
"",
"In this paper the theoretical and practical implications of dropping-from the basic Bayesian coherence principles- the assumption of comparability of every pair of acts is examined. The resulting theory is shown to be still perfectly coherent and has Bayesian theory as a particular case. In particular we question the need of weakening or ruling out some of the axioms that constitute the coherence principles; what are their practical implications; how this drive to the notion of partial information or partial uncertainty in a certain sense; how this partial information is combined with sample information and how this relates to Bayesian methods. We also point out the relation of this approach to rational behaviour with the more (and apparently unrelated) general notion of domination structures as applied to multicrieria decision making.",
"Part 1 Reasoning and behaviour: interpretations of probability beliefs and behaviour inference and decision reasoning and rationality assessment strategies survey of related work. Part 2 Coherent previsions: possibility probability currency upper and lower previsions avoiding sure loss coherence basic properties of coherent previsions coherent probabilities linear previsions and additive probabilities examples of coherent previsions interpretations of prevision and probability objections to behavioural theories of probability. Part 3 Extensions, envelopes and decisions: natural extension extension from a field lower envelopes of linear previsions linear extension invariant linear previsions compactness and extreme points of M(P) desirability and preference equivalent models for beliefs decision making. Part 4 Assessment and elicitation: a general elicitation procedure finitely-generated models and simplex representations steps in assessment process classificatory probability comparative probability other types of assessment. Part 5 The importance of imprecision: uncertainty, indeterminacy and imprecision sources from imprecision information from Bernoulli trials prior-data conflict Bayesian noninformative prioirs indecision axioms of precision practical reasons for precision Bayesian sensitivity analysis second-order probabilities fuzzy sets maximum entropy the Dempster-Shafer theory of belief functions. Part 6 Conditional previsions: updated and contingent previsions separate coherence coherence with unconditional previsions the generalized Bayes rule coherence axioms examples of conditional previsions extension of conditional and marginal previsions conglomerability countable additivity conditioning on events of probability zero updating beliefs. Part 7 Coherent statistical models: general concepts of coherence sampling models coherence of sampling model and posterior previsions inferences from improper priors confidence intervals and relevant subsets proper prior previsions standard Bayesian inference inferences from imprecise priors joint prior previsions. Part 8 Statistical reasoning: a general theroy of natural extension extension to prior previsions extension to predictive previsions extension to posterior previsions posteriors for imprecise sampling models the likelihood principle. Part 9 Structural judgements: independent events independent experiments constructing joint previsions from independent marginals permutability exchangeability Robust Bernoulli models structural judgements. Appendices: verifying coherence N-coherence win and place betting on horses topological structure and L and P separating hyperplane theorems desirability upper and lower variances operational measuremnet procedures the World Cup football experiment regular extension W-coherence."
]
} |
1705.05571 | 2963708149 | Let K be a field equipped with a valuation. Tropical varieties over K can be defined with a theory of Gr "o bner bases taking into account the valuation of K. While generalizing the classical theory of Gr "o bner bases, it is not clear how modern algorithms for computing Gr "o bner bases can be adapted to the tropical case. Among them, one of the most efficient is the celebrated F5 Algorithm of Faug e re. In this article, we prove that, for homogeneous ideals, it can be adapted to the tropical case. We prove termination and correctness. Because of the use of the valuation, the theory of tropical Gr "o b-ner bases is promising for stable computations over polynomial rings over a p-adic field. We provide numerical examples to illustrate time-complexity and p-adic stability of this tropical F5 algorithm. | The computation of tropical varieties over @math with trivial valuation is available in the Gfan package by Anders Jensen (see @cite_6 ), by using standard Gröbner bases computations. Yet, for computation of tropical varieties over general fields, with non-trivial valuation, such techniques are not readily available. Then Chan and Maclagan have developed in @cite_13 a way to extend the theory of Gröbner bases to take into account the valuation and allow tropical computations. Their theory of tropical Gröbner bases is effective and allows, with a suitable division algorithm, a Buchberger algorithm. Following their work, a Matrix-F5 algorithm has been proposed in @cite_11 . | {
"cite_N": [
"@cite_13",
"@cite_6",
"@cite_11"
],
"mid": [
"1588383560",
"",
"2122327334"
],
"abstract": [
"Let K be a field with a valuation and let S be the polynomial ring S:= K[x_1,..., x_n]. We discuss the extension of Groebner theory to ideals in S, taking the valuations of coefficients into account, and describe the Buchberger algorithm in this context. In addition we discuss some implementation and complexity issues. The main motivation comes from tropical geometry, as tropical varieties can be defined using these Groebner bases, but we also give examples showing that the resulting Groebner bases can be substantially smaller than traditional Groebner bases. In the case K = Q with the p-adic valuation the algorithms have been implemented in a Macaulay 2 package.",
"",
"Let K be a field equipped with a valuation. Tropical varieties over K can be defined with a theory of Grobner bases taking into account the valuation of K. Because of the use of the valuation, this theory is promising for stable computations over polynomial rings over a p-adic fields. We design a strategy to compute such tropical Grobner bases by adapting the Matrix-F5 algorithm. Two variants of the Matrix-F5 algorithm, depending on how the Macaulay matrices are built, are available to tropical computation with respective modifications. The former is more numerically stable while the latter is faster. Our study is performed both over any exact field with valuation and some inexact fields like Qp or Fq [[t]]. In the latter case, we track the loss in precision, and show that the numerical stability can compare very favorably to the case of classical Grobner bases when the valuation is non-trivial. Numerical examples are provided."
]
} |
1705.05665 | 2616817002 | Inferring the relations between two images is an important class of tasks in computer vision. Examples of such tasks include computing optical flow and stereo disparity. We treat the relation inference tasks as a machine learning problem and tackle it with neural networks. A key to the problem is learning a representation of relations. We propose a new neural network module, contrast association unit (CAU), which explicitly models the relations between two sets of input variables. Due to the non-negativity of the weights in CAU, we adopt a multiplicative update algorithm for learning these weights. Experiments show that neural networks with CAUs are more effective in learning five fundamental image transformations than conventional neural networks. | A classic model related to the proposed CAU is the energy model for motion detection @cite_29 and stereo disparity @cite_33 . Each unit of the energy model computes the sum of squares of two Gabor filter outputs. No learning is involved in the energy model. There are models which compute the sum of squares of learnable filter outputs such as adaptive-subspace self-organized maps (ASSOM) @cite_24 and independent subspace analysis (ISA) @cite_19 . Similar to CAU, ASSOM also has a competition mechanism (WTA). However, the goal of both ASSOM and ISA is to learn appearance features which are invariant of image transformations, as different from our goal. There is a line of research on relation learning based on Boltzmann machines @cite_4 @cite_31 @cite_25 . While Boltzmann machines allow a probabilistic formulation of relation inference, the training of Boltzmann machines is much more expensive compared to their non-probabilistic counterpart. Non-negative weights have appeared in sum-product networks @cite_20 , multi-layer perceptrons @cite_14 and natural image statistics models @cite_8 . Our derivation of the low-rank approximation of CAUs follows from the low-rank approximation of bilinear units @cite_4 @cite_28 . | {
"cite_N": [
"@cite_14",
"@cite_4",
"@cite_33",
"@cite_8",
"@cite_28",
"@cite_29",
"@cite_24",
"@cite_19",
"@cite_31",
"@cite_25",
"@cite_20"
],
"mid": [
"2093522043",
"2136163184",
"1964291102",
"2022930567",
"2532034655",
"2108992228",
"2021175324",
"2124486835",
"2020315425",
"2198382652",
"2949869425"
],
"abstract": [
"People can understand complex structures if they relate to more isolated yet understandable concepts. Despite this fact, popular pattern recognition tools, such as decision tree or production rule learners, produce only flat models which do not build intermediate data representations. On the other hand, neural networks typically learn hierarchical but opaque models. We show how constraining neurons' weights to be nonnegative improves the interpretability of a network's operation. We analyze the proposed method on large data sets: the MNIST digit recognition data and the Reuters text categorization data. The patterns learned by traditional and constrained network are contrasted to those learned with principal component analysis and nonnegative matrix factorization.",
"To allow the hidden units of a restricted Boltzmann machine to model the transformation between two successive images, Memisevic and Hinton (2007) introduced three-way multiplicative interactions that use the intensity of a pixel in the first image as a multiplicative gain on a learned, symmetric weight between a pixel in the second image and a hidden unit. This creates cubically many parameters, which form a three-dimensional interaction tensor. We describe a low-rank approximation to this interaction tensor that uses a sum of factors, each of which is a three-way outer product. This approximation allows efficient learning of transformations between larger image patches. Since each factor can be viewed as an image filter, the model as a whole learns optimal filter pairs for efficiently representing transformations. We demonstrate the learning of optimal filter pairs from various synthetic and real image sequences. We also show how learning about image transformations allows the model to perform a simple visual analogy task, and we show how a completely unsupervised network trained on transformations perceives multiple motions of transparent dot patterns in the same way as humans.",
"Neurophysiological data support two models for the disparity selectivity of binocular simple and complex cells in primary visual cortex. These involve binocular combinations of monocular receptive fields that are shifted in retinal position (the position-shift model) or in phase (the phase-shift model) between the two eyes. This article presents a formal description and analysis of a binocular energy model with these forms of disparity selectivity. We propose how one might measure the relative contributions of phase and position shifts in simple and complex cells. The analysis also reveals ambiguities in disparity encoding that are inherent in these model neurons, suggesting a need for a second stage of processing. We propose that linear pooling of the binocular responses across orientations and scales (spatial frequency) is capable of producing an unambiguous representation of disparity.",
"Abstract An important property of visual systems is to be simultaneously both selective to specific patterns found in the sensory input and invariant to possible variations. Selectivity and invariance (tolerance) are opposing requirements. It has been suggested that they could be joined by iterating a sequence of elementary selectivity and tolerance computations. It is, however, unknown what should be selected or tolerated at each level of the hierarchy. We approach this issue by learning the computations from natural images. We propose and estimate a probabilistic model of natural images that consists of three processing layers. Two natural image data sets are considered: image patches, and complete visual scenes downsampled to the size of small patches. For both data sets, we find that in the first two layers, simple and complex cell-like computations are performed. In the third layer, we mainly find selectivity to longer contours; for patch data, we further find some selectivity to texture, while for the downsampled complete scenes, some selectivity to curvature is observed.",
"Bilinear models provide rich representations compared with linear models. They have been applied in various visual tasks, such as object recognition, segmentation, and visual question-answering, to get state-of-the-art performances taking advantage of the expanded representations. However, bilinear representations tend to be high-dimensional, limiting the applicability to computationally complex tasks. We propose low-rank bilinear pooling using Hadamard product for an efficient attention mechanism of multimodal learning. We show that our model outperforms compact bilinear pooling in visual question-answering tasks with the state-of-the-art results on the VQA dataset, having a better parsimonious property.",
"A motion sequence may be represented as a single pattern in x–y–t space; a velocity of motion corresponds to a three-dimensional orientation in this space. Motion sinformation can be extracted by a system that responds to the oriented spatiotemporal energy. We discuss a class of models for human motion mechanisms in which the first stage consists of linear filters that are oriented in space-time and tuned in spatial frequency. The outputs of quadrature pairs of such filters are squared and summed to give a measure of motion energy. These responses are then fed into an opponent stage. Energy models can be built from elements that are consistent with known physiology and psychophysics, and they permit a qualitative understanding of a variety of motion phenomena.",
"A new self-organizing map (SOM) architecture called the ASSOM (adaptive-subspace SOM) is shown to create sets of translation-invariant filters when randomly displaced or moving input patterns are used as training data. No analytical functional forms for these filters are thereby postulated. Different kinds of filters are formed by the ASSOM when pictures are rotated during learning, or when they are zoomed. The ASSOM can thus act as a learning feature-extraction stage for pattern recognizers, being able to adapt to many sensory environments and to many different transformation groups of patterns.",
"Olshausen and Field (1996) applied the principle of independence maximization by sparse coding to extract features from natural images. This leads to the emergence of oriented linear filters that have simultaneous localization in space and in frequency, thus resembling Gabor functions and simple cell receptive fields. In this article, we show that the same principle of independence maximization can explain the emergence of phase- and shift-invariant features, similar to those found in complex cells. This new kind of emergence is obtained by maximizing the independence between norms of projections on linear subspaces (instead of the independence of simple linear filter outputs). The norms of the projections on such “independent feature subspaces” then indicate the values of invariant features.",
"We describe a generative model of the relationship between two images. The model is defined as a factored three-way Boltzmann machine, in which hidden variables collaborate to define the joint correlation matrix for image pairs. Modeling the joint distribution over pairs makes it possible to efficiently match images that are the same according to a learned measure of similarity. We apply the model to several face matching tasks, and show that it learns to represent the input images using task-specific basis functions. Matching performance is superior to previous similar generative models, including recent conditional models of transformations. We also show that the model can be used as a plug-in matching score to perform invariant classification.",
"Relation learning is a fundamental operation in many computer vision tasks. Recently, high-order Boltzmann machine and its variants have exhibited the great power of modelling various data relation. However, most of them are unsupervised learning models which are not very discriminative and thus cannot server as a standalone solution to relation learning tasks. In this paper, we explore supervised learning algorithms and propose a new model named Conditional High-order Boltzmann Machine (CHBM), which can be directly used as a bilinear classifier to assign similarity scores for pairwise images. Then, to better deal with complex data relation, we propose a gated version of CHBM which untangles factors of variation by exploiting a set of latent variables to gate classification. We perform four-order tensor factorization for parameter reduction, and present two efficient supervised learning algorithms from the perspectives of being generative and discriminative, respectively. The experimental results of image transformation visualization, binary-way classification and face verification demonstrate that, by performing supervised learning, our models can greatly improve the performance.",
"The key limiting factor in graphical model inference and learning is the complexity of the partition function. We thus ask the question: what are general conditions under which the partition function is tractable? The answer leads to a new kind of deep architecture, which we call sum-product networks (SPNs). SPNs are directed acyclic graphs with variables as leaves, sums and products as internal nodes, and weighted edges. We show that if an SPN is complete and consistent it represents the partition function and all marginals of some graphical model, and give semantics to its nodes. Essentially all tractable graphical models can be cast as SPNs, but SPNs are also strictly more general. We then propose learning algorithms for SPNs, based on backpropagation and EM. Experiments show that inference and learning with SPNs can be both faster and more accurate than with standard deep networks. For example, SPNs perform image completion better than state-of-the-art deep networks for this task. SPNs also have intriguing potential connections to the architecture of the cortex."
]
} |
1705.05311 | 2951471313 | A significant part of the largest Knowledge Graph today, the Linked Open Data cloud, consists of metadata about documents such as publications, news reports, and other media articles. While the widespread access to the document metadata is a tremendous advancement, it is yet not so easy to assign semantic annotations and organize the documents along semantic concepts. Providing semantic annotations like concepts in SKOS thesauri is a classical research topic, but typically it is conducted on the full-text of the documents. For the first time, we offer a systematic comparison of classification approaches to investigate how far semantic annotations can be conducted using just the metadata of the documents such as titles published as labels on the Linked Open Data cloud. We compare the classifications obtained from analyzing the documents' titles with semantic annotations obtained from analyzing the full-text. Apart from the prominent text classification baselines kNN and SVM, we also compare recent techniques of Learning to Rank and neural networks and revisit the traditional methods logistic regression, Rocchio, and Naive Bayes. The results show that across three of our four datasets, the performance of the classifications using only titles reaches over 90 of the quality compared to the classification performance when using the full-text. Thus, conducting document classification by just using the titles is a reasonable approach for automated semantic annotation and opens up new possibilities for enriching Knowledge Graphs. | Most earlier work on the multi-label classification task with many possible output labels relies on nearest neighbor searches (kNN). Using the union of labels as well as separately voting for each individual label among neighbors is a common choice in these nearest neighbor-based classifiers @cite_2 @cite_34 @cite_13 @cite_3 @cite_19 . Concept extraction @cite_15 refers to explicitly finding known concept-specific phrases in the documents. The extracted concepts are re-weighted by inverse document frequency, as in the well-known TF-IDF @cite_28 retrieval model. In our prior work @cite_5 , we have conducted an exhaustive comparison of concept extraction and feature re-weighting methods using kNN as a multi-label classifier. | {
"cite_N": [
"@cite_28",
"@cite_3",
"@cite_19",
"@cite_2",
"@cite_5",
"@cite_15",
"@cite_34",
"@cite_13"
],
"mid": [
"1978394996",
"",
"",
"2052684427",
"2538769922",
"2068903979",
"",
"2146241755"
],
"abstract": [
"The experimental evidence accumulated over the past 20 years indicates that textindexing systems based on the assignment of appropriately weighted single terms produce retrieval results that are superior to those obtainable with other more elaborate text representations. These results depend crucially on the choice of effective term weighting systems. This paper summarizes the insights gained in automatic term weighting, and provides baseline single term indexing models with which other more elaborate content analysis procedures can be compared.",
"",
"",
"Multi-label learning originated from the investigation of text categorization problem, where each document may belong to several predefined topics simultaneously. In multi-label learning, the training set is composed of instances each associated with a set of labels, and the task is to predict the label sets of unseen instances through analyzing training instances with known label sets. In this paper, a multi-label lazy learning approach named ML-KNN is presented, which is derived from the traditional K-nearest neighbor (KNN) algorithm. In detail, for each unseen instance, its K nearest neighbors in the training set are firstly identified. After that, based on statistical information gained from the label sets of these neighboring instances, i.e. the number of neighboring instances belonging to each possible class, maximum a posteriori (MAP) principle is utilized to determine the label set for the unseen instance. Experiments on three different real-world multi-label learning problems, i.e. Yeast gene functional analysis, natural scene classification and automatic web page categorization, show that ML-KNN achieves superior performance to some well-established multi-label learning algorithms.",
"Extracting parts of a text document relevant to a class label is a critical information retrieval task. We propose a semi-supervised multi-label topic model for jointly achieving document and sentence-level class inferences. Under our model, each sentence is associated with only a subset of the document's labels (including possibly none of them), with the label set of the document the union of the labels of all of its sentences. For training, we use both labeled documents, and, typically, a larger set of unlabeled documents. Our model, in a semisupervised fashion, discovers the topics present, learns associations between topics and class labels, predicts labels for new (or unlabeled) documents, and determines label associations for each sentence in every document. For learning, our model does not require any ground-truth labels on sentences. We develop a Hamiltonian Monte Carlo based algorithm for efficiently sampling from the joint label distribution over all sentences, a very high-dimensional discrete space. Our experiments show that our approach outperforms several benchmark methods with respect to both document and sentence-level classification, as well as test set log-likelihood. All code for replicating our experiments is available from https: github.com hsoleimani MLTM.",
"When recommending news items, most of the traditional algorithms are based on TF-IDF, i.e., a term-based weighting method which is mostly used in information retrieval and text mining. However, many new technologies have been made available since the introduction of TF-IDF. This paper proposes a new method for recommending news items based on TF-IDF and a domain ontology. It is demonstrated that adapting TF-IDF with the semantics of a domain ontology, resulting in Concept Frequency - Inverse Document Frequency (CF-IDF), yields better results than using the original TF-IDF method. CF-IDF is built and tested in Athena, a recommender extension to the Hermes news personalization framework. Athena employs a user profile to store concepts or terms found in news items browsed by the user. The framework recommends new articles to the user using a traditional TF-IDF recommender and the CF-IDF recommender. A statistical evaluation of both methods shows that the use of an ontology significantly improves the performance of a traditional recommender.",
"",
"Multi-label classification methods are increasingly required by modern applications, such as protein function classification, music categorization, and semantic scene classification. This article introduces the task of multi-label classification, organizes the sparse related literature into a structured presentation and performs comparative experimental results of certain multilabel classification methods. It also contributes the definition of concepts for the quantification of the multi-label nature of a data set."
]
} |
1705.05311 | 2951471313 | A significant part of the largest Knowledge Graph today, the Linked Open Data cloud, consists of metadata about documents such as publications, news reports, and other media articles. While the widespread access to the document metadata is a tremendous advancement, it is yet not so easy to assign semantic annotations and organize the documents along semantic concepts. Providing semantic annotations like concepts in SKOS thesauri is a classical research topic, but typically it is conducted on the full-text of the documents. For the first time, we offer a systematic comparison of classification approaches to investigate how far semantic annotations can be conducted using just the metadata of the documents such as titles published as labels on the Linked Open Data cloud. We compare the classifications obtained from analyzing the documents' titles with semantic annotations obtained from analyzing the full-text. Apart from the prominent text classification baselines kNN and SVM, we also compare recent techniques of Learning to Rank and neural networks and revisit the traditional methods logistic regression, Rocchio, and Naive Bayes. The results show that across three of our four datasets, the performance of the classifications using only titles reaches over 90 of the quality compared to the classification performance when using the full-text. Thus, conducting document classification by just using the titles is a reasonable approach for automated semantic annotation and opens up new possibilities for enriching Knowledge Graphs. | In the biomedical domain, the most popular approach is Learning to Rank @cite_20 @cite_23 . The algorithm learns a ranking of the MeSH terms. In multi-label classification, however, a hard decision is necessary to enable fully automated classification. Thus, Learning to Rank is typically adjusted for multi-labeling by imposing a hard cut-off. There are also approaches that use Learning to Rank along with dynamic cut-off techniques @cite_10 . The most prominent approach to adapt classifiers for multi-labeling is binary relevance @cite_36 @cite_13 . Other options include the chaining @cite_32 as well as stacking @cite_12 @cite_29 of classifiers. While the former is not well-suited for high amounts of considered labels, we also include a variation of the latter idea in our comparison. Bi and Kwok @cite_31 approach the multi-label classification task from a different direction. They strive for more efficient multi-label classification and proper treatment of label correlation by transforming the label indicator matrix. | {
"cite_N": [
"@cite_13",
"@cite_36",
"@cite_29",
"@cite_32",
"@cite_23",
"@cite_31",
"@cite_10",
"@cite_12",
"@cite_20"
],
"mid": [
"2146241755",
"",
"",
"1999954155",
"2428528690",
"1546425806",
"2606900069",
"1863422369",
"2152143870"
],
"abstract": [
"Multi-label classification methods are increasingly required by modern applications, such as protein function classification, music categorization, and semantic scene classification. This article introduces the task of multi-label classification, organizes the sparse related literature into a structured presentation and performs comparative experimental results of certain multilabel classification methods. It also contributes the definition of concepts for the quantification of the multi-label nature of a data set.",
"",
"",
"The widely known binary relevance method for multi-label classification, which considers each label as an independent binary problem, has often been overlooked in the literature due to the perceived inadequacy of not directly modelling label correlations. Most current methods invest considerable complexity to model interdependencies between labels. This paper shows that binary relevance-based methods have much to offer, and that high predictive performance can be obtained without impeding scalability to large datasets. We exemplify this with a novel classifier chains method that can model label correlations while maintaining acceptable computational complexity. We extend this approach further in an ensemble framework. An extensive empirical evaluation covers a broad range of multi-label datasets with a variety of evaluation metrics. The results illustrate the competitiveness of the chaining method against related and state-of-the-art methods, both in terms of predictive performance and time complexity.",
"Motivation: Medical Subject Headings (MeSH) indexing, which is to assign a set of MeSH main headings to citations, is crucial for many important tasks in biomedical text mining and information retrieval. Large-scale MeSH indexing has two challenging aspects: the citation side and MeSH side. For the citation side, all existing methods, including Medical Text Indexer (MTI) by National Library of Medicine and the state-of-the-art method, MeSHLabeler, deal with text by bag-of-words, which cannot capture semantic and context-dependent information well. Methods: We propose DeepMeSH that incorporates deep semantic information for large-scale MeSH indexing. It addresses the two challenges in both citation and MeSH sides. The citation side challenge is solved by a new deep semantic representation, D2V-TFIDF, which concatenates both sparse and dense semantic representations. The MeSH side challenge is solved by using the ‘learning to rank’ framework of MeSHLabeler, which integrates various types of evidence generated from the new semantic representation. Results: DeepMeSH achieved a Micro F-measure of 0.6323, 2 higher than 0.6218 of MeSHLabeler and 12 higher than 0.5637 of MTI, for BioASQ3 challenge data with 6000 citations. Availability and Implementation: The software is available upon request. Contact: nc.ude.naduf@fsuhz Supplementary information: Supplementary data are available at Bioinformatics online.",
"In multi-label classification, each sample can be associated with a set of class labels. When the number of labels grows to the hundreds or even thousands, existing multi-label classification methods often become computationally inefficient. In recent years, a number of remedies have been proposed. However, they are based either on simple dimension reduction techniques or involve expensive optimization problems. In this paper, we address this problem by selecting a small subset of class labels that can approximately span the original label space. This is performed by an efficient randomized sampling procedure where the sampling probability of each class label re flects its importance among all the labels. Experiments on a number of real-world multi-label data sets with many labels demonstrate the appealing performance and efficiency of the proposed algorithm.",
"Background MeSH indexing is the task of assigning relevant MeSH terms based on a manual reading of scholarly publications by human indexers. The task is highly important for improving literature retrieval and many other scientific investigations in biomedical research. Unfortunately, given its manual nature, the process of MeSH indexing is both time-consuming (new articles are not immediately indexed until 2 or 3 months later) and costly (approximately ten dollars per article). In response, automatic indexing by computers has been previously proposed and attempted but remains challenging. In order to advance the state of the art in automatic MeSH indexing, a community-wide shared task called BioASQ was recently organized.",
"We introduce a new stacking-like approach for multi-value classification. We apply this classification scheme using Naive Bayes, Rocchio and kNN classifiers on the well-known Reuters dataset. We use part-of-speech tagging for stopword removal. We show that our setup performs almost as well as other approaches that use the full article text even though we only classify headlines. Finally, we apply a Rocchio classifier on a dataset from a Web 2.0 site and show that it is suitable for semi-automated labelling (often called tagging) of short texts and is faster than other approaches.",
"Background Due to the high cost of manual curation of key aspects from the scientific literature, automated methods for assisting this process are greatly desired. Here, we report a novel approach to facilitate MeSH indexing, a challenging task of assigning MeSH terms to MEDLINE citations for their archiving and retrieval. @PARASPLIT Methods Unlike previous methods for automatic MeSH term assignment, we reformulate the indexing task as a ranking problem such that relevant MeSH headings are ranked higher than those irrelevant ones. Specifically, for each document we retrieve 20 neighbor documents, obtain a list of MeSH main headings from neighbors, and rank the MeSH main headings using ListNet–a learning-to-rank algorithm. We trained our algorithm on 200 documents and tested on a previously used benchmark set of 200 documents and a larger dataset of 1000 documents. @PARASPLIT Results Tested on the benchmark dataset, our method achieved a precision of 0.390, recall of 0.712, and mean average precision (MAP) of 0.626. In comparison to the state of the art, we observe statistically significant improvements as large as 39 in MAP (p-value <0.001). Similar significant improvements were also obtained on the larger document set. @PARASPLIT Conclusion Experimental results show that our approach makes the most accurate MeSH predictions to date, which suggests its great potential in making a practical impact on MeSH indexing. Furthermore, as discussed the proposed learning framework is robust and can be adapted to many other similar tasks beyond MeSH indexing in the biomedical domain. All data sets are available at: ."
]
} |
1705.05311 | 2951471313 | A significant part of the largest Knowledge Graph today, the Linked Open Data cloud, consists of metadata about documents such as publications, news reports, and other media articles. While the widespread access to the document metadata is a tremendous advancement, it is yet not so easy to assign semantic annotations and organize the documents along semantic concepts. Providing semantic annotations like concepts in SKOS thesauri is a classical research topic, but typically it is conducted on the full-text of the documents. For the first time, we offer a systematic comparison of classification approaches to investigate how far semantic annotations can be conducted using just the metadata of the documents such as titles published as labels on the Linked Open Data cloud. We compare the classifications obtained from analyzing the documents' titles with semantic annotations obtained from analyzing the full-text. Apart from the prominent text classification baselines kNN and SVM, we also compare recent techniques of Learning to Rank and neural networks and revisit the traditional methods logistic regression, Rocchio, and Naive Bayes. The results show that across three of our four datasets, the performance of the classifications using only titles reaches over 90 of the quality compared to the classification performance when using the full-text. Thus, conducting document classification by just using the titles is a reasonable approach for automated semantic annotation and opens up new possibilities for enriching Knowledge Graphs. | Zhang and Zhou @cite_1 have proposed to train a separate neural network for each label along with a dedicated loss function. However, this approach does not scale to high amounts of possible output labels. One year later, the same authors suggest a lazy-learning multi-label variant of kNN @cite_2 , which is considered in our comparison. Nam al @cite_27 adapt fully connected feed-forward neural networks for multi-label classification by learning a threshold that determines whether a label should be assigned or not. | {
"cite_N": [
"@cite_27",
"@cite_1",
"@cite_2"
],
"mid": [
"2951829787",
"2119466907",
"2052684427"
],
"abstract": [
"Neural networks have recently been proposed for multi-label classification because they are able to capture and model label dependencies in the output layer. In this work, we investigate limitations of BP-MLL, a neural network (NN) architecture that aims at minimizing pairwise ranking error. Instead, we propose to use a comparably simple NN approach with recently proposed learning techniques for large-scale multi-label text classification tasks. In particular, we show that BP-MLL's ranking loss minimization can be efficiently and effectively replaced with the commonly used cross entropy error function, and demonstrate that several advances in neural network training that have been developed in the realm of deep learning can be effectively employed in this setting. Our experimental results show that simple NN models equipped with advanced techniques such as rectified linear units, dropout, and AdaGrad perform as well as or even outperform state-of-the-art approaches on six large-scale textual datasets with diverse characteristics.",
"In multilabel learning, each instance in the training set is associated with a set of labels and the task is to output a label set whose size is unknown a priori for each unseen instance. In this paper, this problem is addressed in the way that a neural network algorithm named BP-MLL, i.e., backpropagation for multilabel learning, is proposed. It is derived from the popular backpropagation algorithm through employing a novel error function capturing the characteristics of multilabel learning, i.e., the labels belonging to an instance should be ranked higher than those not belonging to that instance. Applications to two real-world multilabel learning problems, i.e., functional genomics and text categorization, show that the performance of BP-MLL is superior to that of some well-established multilabel learning algorithms",
"Multi-label learning originated from the investigation of text categorization problem, where each document may belong to several predefined topics simultaneously. In multi-label learning, the training set is composed of instances each associated with a set of labels, and the task is to predict the label sets of unseen instances through analyzing training instances with known label sets. In this paper, a multi-label lazy learning approach named ML-KNN is presented, which is derived from the traditional K-nearest neighbor (KNN) algorithm. In detail, for each unseen instance, its K nearest neighbors in the training set are firstly identified. After that, based on statistical information gained from the label sets of these neighboring instances, i.e. the number of neighboring instances belonging to each possible class, maximum a posteriori (MAP) principle is utilized to determine the label set for the unseen instance. Experiments on three different real-world multi-label learning problems, i.e. Yeast gene functional analysis, natural scene classification and automatic web page categorization, show that ML-KNN achieves superior performance to some well-established multi-label learning algorithms."
]
} |
1705.05344 | 2615735215 | As we aim to control complex systems, use of a simulator in model-based reinforcement learning is becoming more common. However, it has been challenging to overcome the Reality Gap, which comes from nonlinear model bias and susceptibility to disturbance. To address these problems, we propose a novel algorithm that combines data-driven system identification approach (Gaussian Process) with a Differential-Dynamic-Programming-based robust optimal control method (Iterative Linear Quadratic Control). Our algorithm uses the simulator's model as the mean function for a Gaussian Process and learns only the difference between the simulator's prediction and actual observations, making it a natural hybrid of simulation and real-world observation. We show that our approach quickly corrects incorrect models, comes up with robust optimal controllers, and transfers its acquired model knowledge to new tasks efficiently. | Most model-based reinforcement learning has both model learning (system identification) and policy optimization components @cite_9 . The data for a model comes either from real world or simulation, and is combined to construct a model via nonlinear function approximators such as Locally Weighted Regression @cite_15 , Gaussian Processes @cite_18 , or Neural Networks @cite_27 . Once the model is built, a typical policy gradient method computes the derivatives of the cost function with respect to control parameters @cite_17 @cite_25 . | {
"cite_N": [
"@cite_18",
"@cite_9",
"@cite_27",
"@cite_15",
"@cite_25",
"@cite_17"
],
"mid": [
"2107386393",
"1977655452",
"2060329955",
"1689445748",
"",
"2140135625"
],
"abstract": [
"The GPML toolbox provides a wide range of functionality for Gaussian process (GP) inference and prediction. GPs are specified by mean and covariance functions; we offer a library of simple mean and covariance functions and mechanisms to compose more complex ones. Several likelihood functions are supported including Gaussian and heavy-tailed for regression as well as others suitable for classification. Finally, a range of inference methods is provided, including exact and variational inference, Expectation Propagation, and Laplace's method dealing with non-Gaussian likelihoods and FITC for dealing with large regression tasks.",
"Reinforcement learning offers to robotics a framework and set of tools for the design of sophisticated and hard-to-engineer behaviors. Conversely, the challenges of robotic problems provide both inspiration, impact, and validation for developments in reinforcement learning. The relationship between disciplines has sufficient promise to be likened to that between physics and mathematics. In this article, we attempt to strengthen the links between the two research communities by providing a survey of work in reinforcement learning for behavior generation in robots. We highlight both key challenges in robot reinforcement learning as well as notable successes. We discuss how contributions tamed the complexity of the domain and study the role of algorithms, representations, and prior knowledge in achieving these successes. As a result, a particular focus of our paper lies on the choice between model-based and model-free as well as between value-function-based and policy-search methods. By analyzing a simple problem in some detail we demonstrate how reinforcement learning approaches may be profitably applied, and we note throughout open questions and the tremendous potential for future research.",
"Currently, the field of sensory-motor neuroscience lacks a computational model that can replicate real-time control of biological brain. Due to incomplete neural and anatomical data, traditional neural network training methods fail to model the sensory-motor systems. Here we introduce a novel modeling method based on stochastic optimal control framework which is well suited for this purpose. Our controller is implemented with a recurrent neural network (RNN) whose goal is approximating the optimal global control law for the given plant and cost function. We employ a risk-sensitive objective function proposed by Jacobson (1973) for robustness of controller. For maximum optimization efficiency, we introduce a step response sampling method, which minimizes complexity of the optimization problem. We use conjugate gradient descent method for optimization, and gradient is calculated via Pontryagins maximum principle. In the end, we obtain highly stable and robust RNN controllers that can generate infinite varieties of attractor dynamics of the plant, which are proposed as building blocks of movement generation. We show two such examples, a point attractor based and a limit-cycle based dynamics.",
"This paper surveys locally weighted learning, a form of lazy learning and memory-based learning, and focuses on locally weighted linear regression. The survey discusses distance functions, smoothing parameters, weighting functions, local model structures, regularization of the estimates and bias, assessing predictions, handling noisy data and outliers, improving the quality of predictions by tuning fit parameters, interference between old and new data, implementing locally weighted learning efficiently, and applications of locally weighted learning. A companion paper surveys how locally weighted learning can be used in robot learning and control.",
"",
"In this paper, we introduce PILCO, a practical, data-efficient model-based policy search method. PILCO reduces model bias, one of the key problems of model-based reinforcement learning, in a principled way. By learning a probabilistic dynamics model and explicitly incorporating model uncertainty into long-term planning, PILCO can cope with very little data and facilitates learning from scratch in only a few trials. Policy evaluation is performed in closed form using state-of-the-art approximate inference. Furthermore, policy gradients are computed analytically for policy improvement. We report unprecedented learning efficiency on challenging and high-dimensional control tasks."
]
} |
1705.05344 | 2615735215 | As we aim to control complex systems, use of a simulator in model-based reinforcement learning is becoming more common. However, it has been challenging to overcome the Reality Gap, which comes from nonlinear model bias and susceptibility to disturbance. To address these problems, we propose a novel algorithm that combines data-driven system identification approach (Gaussian Process) with a Differential-Dynamic-Programming-based robust optimal control method (Iterative Linear Quadratic Control). Our algorithm uses the simulator's model as the mean function for a Gaussian Process and learns only the difference between the simulator's prediction and actual observations, making it a natural hybrid of simulation and real-world observation. We show that our approach quickly corrects incorrect models, comes up with robust optimal controllers, and transfers its acquired model knowledge to new tasks efficiently. | The goal of our algorithm is closely aligned with those of @cite_16 and @cite_28 . @cite_16 has proposed a framework in which the robot maintains two controllers, one for the simulator and another for the real world, and aims to narrow the difference. @cite_28 assumes a deterministic real world, constructs an optimal policy based on the simulator's deterministic model, and evaluates its performance in the real world, while successively augmenting the simulator's model with time-dependent corrections based on real-world observations. Our method considers a stochastic system and the correction is not time-dependent. | {
"cite_N": [
"@cite_28",
"@cite_16"
],
"mid": [
"2132602063",
"31984690"
],
"abstract": [
"In the model-based policy search approach to reinforcement learning (RL), policies are found using a model (or \"simulator\") of the Markov decision process. However, for high-dimensional continuous-state tasks, it can be extremely difficult to build an accurate model, and thus often the algorithm returns a policy that works in simulation but not in real-life. The other extreme, model-free RL, tends to require infeasibly large numbers of real-life trials. In this paper, we present a hybrid algorithm that requires only an approximate model, and only a small number of real-life trials. The key idea is to successively \"ground\" the policy evaluations using real-life trials, but to rely on the approximate model to suggest local changes. Our theoretical results show that this algorithm achieves near-optimal performance in the real system, even when the model is only approximate. Empirical results also demonstrate that---when given only a crude model and a small number of real-life trials---our algorithm can obtain near-optimal performance in the real system.",
"Abstract In this work a new method to evolutionary robotics is proposed, it combines into asingle framework, learning from reality and simulations. An illusory sub-system is incorporated as an integral part of an autonomous system. The adaptation of the illusory system results from minimizing differences of robot behavior evaluations in reality and in simulations. Behavior guides the illusory adaptation by sampling task-relevant instances of the world. Thus explicit calibration is not required. We remark two attributes of the presented methodology: (i) it is a promising approach for crossing the reality-gap among simulation and reality in evolutionary robotics, and (ii) it allows to generate automatically models and theories of the real robot environment expressed as simulations. We present validation experiments on locomotive behavior acquisition for legged robots."
]
} |
1705.05396 | 2615676335 | Probabilistic modeling enables combining domain knowledge with learning from data, thereby supporting learning from fewer training instances than purely data-driven methods. However, learning probabilistic models is difficult and has not achieved the level of performance of methods such as deep neural networks on many tasks. In this paper, we attempt to address this issue by presenting a method for learning the parameters of a probabilistic program using backpropagation. Our approach opens the possibility to building deep probabilistic programming models that are trained in a similar way to neural networks. | Stan @cite_11 is based on reverse-mode automatic differentiation, like ours. However, Stan uses this for Hamiltonian Monte Carlo, whereas we use it for back-propagation. Stan also requires a more restricted form of model than we do. Stan requires that the model be differentiable with respect to the variables in the model. In our framework, the model can be non-differentiable and even discontinuous with respect to the variables. It need only be differentiable with respect to the learnable parameters of the model. This makes it applicable to general-purpose probabilistic programming languages like Church. | {
"cite_N": [
"@cite_11"
],
"mid": [
"2577537660"
],
"abstract": [
"Stan is a probabilistic programming language for specifying statistical models. A Stan program imperatively defines a log probability function over parameters conditioned on specified data and constants. As of version 2.14.0, Stan provides full Bayesian inference for continuous-variable models through Markov chain Monte Carlo methods such as the No-U-Turn sampler, an adaptive form of Hamiltonian Monte Carlo sampling. Penalized maximum likelihood estimates are calculated using optimization methods such as the limited memory Broyden-Fletcher-Goldfarb-Shanno algorithm. Stan is also a platform for computing log densities and their gradients and Hessians, which can be used in alternative algorithms such as variational Bayes, expectation propagation, and marginal inference using approximate integration. To this end, Stan is set up so that the densities, gradients, and Hessians, along with intermediate quantities of the algorithm such as acceptance probabilities, are easily accessible. Stan can be called from the command line using the cmdstan package, through R using the rstan package, and through Python using the pystan package. All three interfaces support sampling and optimization-based inference with diagnostics and posterior analysis. rstan and pystan also provide access to log probabilities, gradients, Hessians, parameter transforms, and specialized plotting."
]
} |
1705.05396 | 2615676335 | Probabilistic modeling enables combining domain knowledge with learning from data, thereby supporting learning from fewer training instances than purely data-driven methods. However, learning probabilistic models is difficult and has not achieved the level of performance of methods such as deep neural networks on many tasks. In this paper, we attempt to address this issue by presenting a method for learning the parameters of a probabilistic program using backpropagation. Our approach opens the possibility to building deep probabilistic programming models that are trained in a similar way to neural networks. | Edward @cite_3 is a probabilistic programming language that enables explicit representation of inference models. These inference models are implemented in Tensor Flow, enabling many of the scalability benefits of neural networks. The main difference between Edward and our approach is that Edward requires the inference model to be written explicitly whereas our approach is black box, with inference being worked out automatically. Having to specify an inference model can be an advantage, because it enables you to encode algorithms (e.g. variational methods) that couldn't be easily derived. However, it might be hard to scale to more complex models and would be particularly difficult for non-experts to use. | {
"cite_N": [
"@cite_3"
],
"mid": [
"2539792571"
],
"abstract": [
"Probabilistic modeling is a powerful approach for analyzing empirical information. We describe Edward, a library for probabilistic modeling. Edward's design reflects an iterative process pioneered by George Box: build a model of a phenomenon, make inferences about the model given data, and criticize the model's fit to the data. Edward supports a broad class of probabilistic models, efficient algorithms for inference, and many techniques for model criticism. The library builds on top of TensorFlow to support distributed training and hardware such as GPUs. Edward enables the development of complex probabilistic models and their algorithms at a massive scale."
]
} |
1705.05396 | 2615676335 | Probabilistic modeling enables combining domain knowledge with learning from data, thereby supporting learning from fewer training instances than purely data-driven methods. However, learning probabilistic models is difficult and has not achieved the level of performance of methods such as deep neural networks on many tasks. In this paper, we attempt to address this issue by presenting a method for learning the parameters of a probabilistic program using backpropagation. Our approach opens the possibility to building deep probabilistic programming models that are trained in a similar way to neural networks. | Le, Beydin, and Wood @cite_13 use neural networks to support inference in probabilistic programs, specifically to help create good proposals for sequential importance sampling. Their work differs from ours in that they do not learn the probabilistic program itself but rather an auxiliary network that assists in inference. | {
"cite_N": [
"@cite_13"
],
"mid": [
"2950999850"
],
"abstract": [
"We introduce a method for using deep neural networks to amortize the cost of inference in models from the family induced by universal probabilistic programming languages, establishing a framework that combines the strengths of probabilistic programming and deep learning methods. We call what we do \"compilation of inference\" because our method transforms a denotational specification of an inference problem in the form of a probabilistic program written in a universal programming language into a trained neural network denoted in a neural network specification language. When at test time this neural network is fed observational data and executed, it performs approximate inference in the original model specified by the probabilistic program. Our training objective and learning procedure are designed to allow the trained neural network to be used as a proposal distribution in a sequential importance sampling inference engine. We illustrate our method on mixture models and Captcha solving and show significant speedups in the efficiency of inference."
]
} |
1705.05396 | 2615676335 | Probabilistic modeling enables combining domain knowledge with learning from data, thereby supporting learning from fewer training instances than purely data-driven methods. However, learning probabilistic models is difficult and has not achieved the level of performance of methods such as deep neural networks on many tasks. In this paper, we attempt to address this issue by presenting a method for learning the parameters of a probabilistic program using backpropagation. Our approach opens the possibility to building deep probabilistic programming models that are trained in a similar way to neural networks. | It is naturally possible to give ordinary neural networks a Bayesian interpretation, as in Bayesian deep learning @cite_6 . A variety of specific forms of Bayesian generative neural network frameworks trainable by backpropagation have been developed, such as deep belief nets @cite_7 , deep Boltzmann machines @cite_12 , deep generative models @cite_8 , and deep generative stochastic networks @cite_9 . In contrast to our approach, each of these frameworks defines a specific kind of neural network with a given structure. In contrast, ours is a general framework that applies to programs in a generic probabilistic programming language. | {
"cite_N": [
"@cite_7",
"@cite_8",
"@cite_9",
"@cite_6",
"@cite_12"
],
"mid": [
"",
"1909320841",
"2951446714",
"2338752163",
"189596042"
],
"abstract": [
"",
"We marry ideas from deep neural networks and approximate Bayesian inference to derive a generalised class of deep, directed generative models, endowed with a new algorithm for scalable inference and learning. Our algorithm introduces a recognition model to represent approximate posterior distributions, and that acts as a stochastic encoder of the data. We develop stochastic back-propagation -- rules for back-propagation through stochastic variables -- and use this to develop an algorithm that allows for joint optimisation of the parameters of both the generative and recognition model. We demonstrate on several real-world data sets that the model generates realistic samples, provides accurate imputations of missing data and is a useful tool for high-dimensional data visualisation.",
"We introduce a novel training principle for probabilistic models that is an alternative to maximum likelihood. The proposed Generative Stochastic Networks (GSN) framework is based on learning the transition operator of a Markov chain whose stationary distribution estimates the data distribution. The transition distribution of the Markov chain is conditional on the previous state, generally involving a small move, so this conditional distribution has fewer dominant modes, being unimodal in the limit of small moves. Thus, it is easier to learn because it is easier to approximate its partition function, more like learning to perform supervised function approximation, with gradients that can be obtained by backprop. We provide theorems that generalize recent work on the probabilistic interpretation of denoising autoencoders and obtain along the way an interesting justification for dependency networks and generalized pseudolikelihood, along with a definition of an appropriate joint distribution and sampling mechanism even when the conditionals are not consistent. GSNs can be used with missing inputs and can be used to sample subsets of variables given the rest. We validate these theoretical results with experiments on two image datasets using an architecture that mimics the Deep Boltzmann Machine Gibbs sampler but allows training to proceed with simple backprop, without the need for layerwise pretraining.",
"While perception tasks such as visual object recognition and text understanding play an important role in human intelligence, the subsequent tasks that involve inference, reasoning and planning require an even higher level of intelligence. The past few years have seen major advances in many perception tasks using deep learning models. For higher-level inference, however, probabilistic graphical models with their Bayesian nature are still more powerful and flexible. To achieve integrated intelligence that involves both perception and inference, it is naturally desirable to tightly integrate deep learning and Bayesian models within a principled probabilistic framework, which we call Bayesian deep learning. In this unified framework, the perception of text or images using deep learning can boost the performance of higher-level inference and in return, the feedback from the inference process is able to enhance the perception of text or images. This survey provides a general introduction to Bayesian deep learning and reviews its recent applications on recommender systems, topic models, and control. In this survey, we also discuss the relationship and differences between Bayesian deep learning and other related topics like Bayesian treatment of neural networks.",
"We present a new learning algorithm for Boltzmann machines that contain many layers of hidden variables. Data-dependent expectations are estimated using a variational approximation that tends to focus on a single mode, and dataindependent expectations are approximated using persistent Markov chains. The use of two quite different techniques for estimating the two types of expectation that enter into the gradient of the log-likelihood makes it practical to learn Boltzmann machines with multiple hidden layers and millions of parameters. The learning can be made more efficient by using a layer-by-layer “pre-training” phase that allows variational inference to be initialized with a single bottomup pass. We present results on the MNIST and NORB datasets showing that deep Boltzmann machines learn good generative models and perform well on handwritten digit and visual object recognition tasks."
]
} |
1705.05491 | 2952782294 | We consider the problem of distributed statistical machine learning in adversarial settings, where some unknown and time-varying subset of working machines may be compromised and behave arbitrarily to prevent an accurate model from being learned. This setting captures the potential adversarial attacks faced by Federated Learning -- a modern machine learning paradigm that is proposed by Google researchers and has been intensively studied for ensuring user privacy. Formally, we focus on a distributed system consisting of a parameter server and @math working machines. Each working machine keeps @math data samples, where @math is the total number of samples. The goal is to collectively learn the underlying true model parameter of dimension @math . In classical batch gradient descent methods, the gradients reported to the server by the working machines are aggregated via simple averaging, which is vulnerable to a single Byzantine failure. In this paper, we propose a Byzantine gradient descent method based on the geometric median of means of the gradients. We show that our method can tolerate @math Byzantine failures, and the parameter estimate converges in @math rounds with an estimation error of @math , hence approaching the optimal error rate @math in the centralized and failure-free setting. The total computational complexity of our algorithm is of @math at each working machine and @math at the central server, and the total communication cost is of @math . We further provide an application of our general results to the linear regression problem. A key challenge arises in the above problem is that Byzantine failures create arbitrary and unspecified dependency among the iterations and the aggregated gradients. We prove that the aggregated gradient converges uniformly to the true gradient function. | The present paper intersects with two main areas of research: statistical machine learning and distributed computing. Most related to our work is @cite_17 that we became aware of when preparing this paper. It also studies distributed optimization in adversarial settings, but the setup is different from ours. In particular, their focus is solving an optimization problem, where all @math working machines have access to a common dataset @math and the goal is to collectively compute the minimizer @math of the average cost @math . Importantly, the dataset @math are assumed to be deterministic. In contrast, we adopt the standard statistical learning framework, where each working machine only has access to its own data samples, which are assumed to be generated from some unknown distribution @math , and the goal is to estimate the optimal model parameter @math that minimizes the true prediction error @math --- as mentioned, characterizing the statistical estimation accuracy is a main focus of ours. Our algorithmic approaches and main results are also significantly different. The almost sure convergence is proved in @cite_17 without an explicit characterization of convergence speed nor the estimation errors. | {
"cite_N": [
"@cite_17"
],
"mid": [
"2593047802"
],
"abstract": [
"The growth of data, the need for scalability and the complexity of models used in modern machine learning calls for distributed implementations. Yet, as of today, distributed machine learning frameworks have largely ignored the possibility of arbitrary (i.e., Byzantine) failures. In this paper, we study the robustness to Byzantine failures at the fundamental level of stochastic gradient descent (SGD), the heart of most machine learning algorithms. Assuming a set of @math workers, up to @math of them being Byzantine, we ask how robust can SGD be, without limiting the dimension, nor the size of the parameter space. We first show that no gradient descent update rule based on a linear combination of the vectors proposed by the workers (i.e, current approaches) tolerates a single Byzantine failure. We then formulate a resilience property of the update rule capturing the basic requirements to guarantee convergence despite @math Byzantine workers. We finally propose Krum, an update rule that satisfies the resilience property aforementioned. For a @math -dimensional learning problem, the time complexity of Krum is @math ."
]
} |
1705.05491 | 2952782294 | We consider the problem of distributed statistical machine learning in adversarial settings, where some unknown and time-varying subset of working machines may be compromised and behave arbitrarily to prevent an accurate model from being learned. This setting captures the potential adversarial attacks faced by Federated Learning -- a modern machine learning paradigm that is proposed by Google researchers and has been intensively studied for ensuring user privacy. Formally, we focus on a distributed system consisting of a parameter server and @math working machines. Each working machine keeps @math data samples, where @math is the total number of samples. The goal is to collectively learn the underlying true model parameter of dimension @math . In classical batch gradient descent methods, the gradients reported to the server by the working machines are aggregated via simple averaging, which is vulnerable to a single Byzantine failure. In this paper, we propose a Byzantine gradient descent method based on the geometric median of means of the gradients. We show that our method can tolerate @math Byzantine failures, and the parameter estimate converges in @math rounds with an estimation error of @math , hence approaching the optimal error rate @math in the centralized and failure-free setting. The total computational complexity of our algorithm is of @math at each working machine and @math at the central server, and the total communication cost is of @math . We further provide an application of our general results to the linear regression problem. A key challenge arises in the above problem is that Byzantine failures create arbitrary and unspecified dependency among the iterations and the aggregated gradients. We prove that the aggregated gradient converges uniformly to the true gradient function. | Our work is also closely related to the literature on robust parameter estimation using geometric median. It is shown in @cite_5 that geometric median has a breakdown point of @math , that is, given a collection of @math vectors in @math , at least @math number of points needs to be corrupted in order to arbitrarily perturb the geometric median. A more quantitative robustness result is recently derived in [Lemma 2.1] minsker2015geometric . The geometric median has been applied to distributed machine learning under the one-shot aggregation framework @cite_2 , under the restrictive assumption that the number of data available in each working machine satisfies @math . While we also apply geometric median-of-mean as a sub-routine, our problem setup, overall algorithms and main results are completely different. | {
"cite_N": [
"@cite_5",
"@cite_2"
],
"mid": [
"2136179595",
"1855339086"
],
"abstract": [
"Finite-sample replacement breakdown points are derived for different types of estimators of multivariate location and covariance matrices. The role of various equivariance properties is illustrated. The breakdown point is related to a measure of performance based on large deviations probabilities. Finally, we show that one-step reweighting preserves the breakdown point.",
"We propose a framework for distributed robust statistical learning on big contaminated data . The Distributed Robust Learning (DRL) framework can reduce the computational time of traditional robust learning methods by several orders of magnitude. We analyze the robustness property of DRL, showing that DRL not only preserves the robustness of the base robust learning method, but also tolerates contaminations on a constant fraction of results from computing nodes (node failures). More precisely, even in presence of the most adversarial outlier distribution over computing nodes, DRL still achieves a breakdown point of at least @math , where @math is the break down point of corresponding centralized algorithm. This is in stark contrast with naive division-and-averaging implementation, which may reduce the breakdown point by a factor of @math when @math computing nodes are used. We then specialize the DRL framework for two concrete cases: distributed robust principal component analysis and distributed robust regression. We demonstrate the efficiency and the robustness advantages of DRL through comprehensive simulations and predicting image tags on a large-scale image set."
]
} |
1705.05491 | 2952782294 | We consider the problem of distributed statistical machine learning in adversarial settings, where some unknown and time-varying subset of working machines may be compromised and behave arbitrarily to prevent an accurate model from being learned. This setting captures the potential adversarial attacks faced by Federated Learning -- a modern machine learning paradigm that is proposed by Google researchers and has been intensively studied for ensuring user privacy. Formally, we focus on a distributed system consisting of a parameter server and @math working machines. Each working machine keeps @math data samples, where @math is the total number of samples. The goal is to collectively learn the underlying true model parameter of dimension @math . In classical batch gradient descent methods, the gradients reported to the server by the working machines are aggregated via simple averaging, which is vulnerable to a single Byzantine failure. In this paper, we propose a Byzantine gradient descent method based on the geometric median of means of the gradients. We show that our method can tolerate @math Byzantine failures, and the parameter estimate converges in @math rounds with an estimation error of @math , hence approaching the optimal error rate @math in the centralized and failure-free setting. The total computational complexity of our algorithm is of @math at each working machine and @math at the central server, and the total communication cost is of @math . We further provide an application of our general results to the linear regression problem. A key challenge arises in the above problem is that Byzantine failures create arbitrary and unspecified dependency among the iterations and the aggregated gradients. We prove that the aggregated gradient converges uniformly to the true gradient function. | On the technical front, a crucial step in our convergence proof is to show the geometric median of means of @math i.i.d. random gradients converges to the underlying gradient function @math uniformly over @math . Our proof builds on several ideas from the empirical process theory, which guarantees uniform convergence of the empirical risk function @math to the population risk @math . However, what we need is the uniform convergence of empirical function @math , as well as its version, to the population gradient function @math . To this end, we use concentration inequalities to first establish point-wise convergence and then boost it to uniform convergence via the celebrated @math -net argument. Similar ideas have been used recently in the work @cite_11 , which studies the stationary points of the empirical risk function. | {
"cite_N": [
"@cite_11"
],
"mid": [
"2514392868"
],
"abstract": [
"Most high-dimensional estimation and prediction methods propose to minimize a cost function (empirical risk) that is written as a sum of losses associated to each data point. In this paper we focus on the case of non-convex losses, which is practically important but still poorly understood. Classical empirical process theory implies uniform convergence of the empirical risk to the population risk. While uniform convergence implies consistency of the resulting M-estimator, it does not ensure that the latter can be computed efficiently. In order to capture the complexity of computing M-estimators, we propose to study the landscape of the empirical risk, namely its stationary points and their properties. We establish uniform convergence of the gradient and Hessian of the empirical risk to their population counterparts, as soon as the number of samples becomes larger than the number of unknown parameters (modulo logarithmic factors). Consequently, good properties of the population risk can be carried to the empirical risk, and we can establish one-to-one correspondence of their stationary points. We demonstrate that in several problems such as non-convex binary classification, robust regression, and Gaussian mixture model, this result implies a complete characterization of the landscape of the empirical risk, and of the convergence properties of descent algorithms. We extend our analysis to the very high-dimensional setting in which the number of parameters exceeds the number of samples, and provide a characterization of the empirical risk landscape under a nearly information-theoretically minimal condition. Namely, if the number of samples exceeds the sparsity of the unknown parameters vector (modulo logarithmic factors), then a suitable uniform convergence result takes place. We apply this result to non-convex binary classification and robust regression in very high-dimension."
]
} |
1705.05646 | 2616359666 | We present the first super-linear lower bounds for natural graph problems in the CONGEST model, answering a long-standing open question. Specifically, we show that any exact computation of a minimum vertex cover or a maximum independent set requires @math rounds in the worst case in the CONGEST model, as well as any algorithm for @math -coloring a graph, where @math is the chromatic number of the graph. We further show that such strong lower bounds are not limited to NP-hard problems, by showing two simple graph problems in P which require a quadratic and near-quadratic number of rounds. Finally, we address the problem of computing an exact solution to weighted all-pairs-shortest-paths (APSP), which arguably may be considered as a candidate for having a super-linear lower bound. We show a simple @math lower bound for this problem, which implies a separation between the weighted and unweighted cases, since the latter is known to have a complexity of @math . We also formally prove that the standard Alice-Bob framework is incapable of providing a super-linear lower bound for exact weighted APSP, whose complexity remains an intriguing open question. | One of the most central problems in graph theory is vertex coloring, which has been extensively studied in the context of distributed computing (see, e.g., @cite_25 @cite_22 @cite_27 @cite_11 @cite_57 @cite_6 @cite_52 @cite_44 @cite_48 @cite_40 @cite_35 @cite_7 @cite_23 @cite_14 @cite_46 @cite_8 @cite_36 and references therein). The special case of finding a @math -coloring, where @math is the maximum degree of a node in the network, has been the focus of many of these studies, but is a problem, which can be solved in much less than a sublinear number of rounds. | {
"cite_N": [
"@cite_35",
"@cite_14",
"@cite_22",
"@cite_7",
"@cite_8",
"@cite_36",
"@cite_48",
"@cite_52",
"@cite_6",
"@cite_57",
"@cite_44",
"@cite_27",
"@cite_40",
"@cite_23",
"@cite_46",
"@cite_25",
"@cite_11"
],
"mid": [
"1976011215",
"2009970369",
"",
"88197230",
"2067661972",
"2169324896",
"",
"2030160771",
"2054910423",
"2136185479",
"",
"2065826935",
"2797253418",
"2110673270",
"2279830512",
"2467514673",
"2751147784"
],
"abstract": [
"During and immediately after their deployment, ad hoc and sensor networks lack an efficient communication scheme rendering even the most basic network coordination problems difficult. Before any reasonable communication can take place, nodes must come up with an initial structure that can serve as a foundation for more sophisticated algorithms. In this paper, we consider the problem of obtaining a vertex coloring as such an initial structure. We propose an algorithm that works in the unstructured radio network model. This model captures the characteristics of newly deployed ad hoc and sensor networks, i.e. asynchronous wake-up, no collision-detection, and scarce knowledge about the network topology. When modeling the network as a graph with bounded independence, our algorithm produces a correct coloring with O(Δ) colors in time O(Δ log n) with high probability, where n and Δ are the number of nodes in the network and the maximum degree, respectively. Also, the number of locally used colors depends only on the local node density. Graphs with bounded independence generalize unit disk graphs as well as many other well-known models for wireless multi-hop networks. They allow us to capture aspects such as obstacles, fading, or irregular signal-propagation.",
"The Lovasz Local Lemma (LLL), introduced by Erdos and Lovasz in 1975, is a powerful tool of the probabilistic method that allows one to prove that a set of n \"bad\" events do not happen with non-zero probability, provided that the events have limited dependence. However, the LLL itself does not suggest how to find a point avoiding all bad events. Since the work of Beck (1991) there has been a sustained effort to find a constructive proof (i.e. an algorithm) for the LLL or weaker versions of it. In a major breakthrough Moser and Tardos (2010) showed that a point avoiding all bad events can be found efficiently. They also proposed a distributed parallel version of their algorithm that requires O(log2 n) rounds of communication in a distributed network. In this paper we provide two new distributed algorithms for the LLL that improve on both the efficiency and simplicity of the Moser-Tardos algorithm. For clarity we express our results in terms of the symmetric LLL though both algorithms deal with the asymmetric version as well. Let p bound the probability of any bad event and d be the maximum degree in the dependency graph of the bad events. When epd2 In many graph coloring problems the existence of a valid coloring is established by one or more applications of the LLL. Using our LLL algorithms, we give logarithmic-time distributed algorithms for frugal coloring, defective coloring, coloring girth-4 (triangle-free) and girth-5 graphs, edge coloring, and list coloring.",
"",
"We deterministically compute a Δ + 1 coloring in time O(Δ5c+2 ċ (Δ5)2 c (Δ1)e + (Δ1)e + log* n) and O(Δ5c+2 ċ (Δ5)1 c Δe + Δe + (Δ5)d log Δ5 log n) for arbitrary constants d, e and arbitrary constant integer c, where Δi is defined as the maximal number of nodes within distance i for a node and Δ := Δ1. Our greedy algorithm improves the state-of-the-art Δ + 1 coloring algorithms for a large class of graphs, e.g. graphs of moderate neighborhood growth. We also state and analyze a randomized coloring algorithm in terms of the chromatic number, the run time and the used colors. If Δ ∈ Ω(log1+1 log* n n) and χ ∈ O(Δ log1+1 log* n n) then our algorithm executes in time O(log χ + log* n) with high probability. For graphs of polylogarithmic chromatic number the analysis reveals an exponential gap compared to the fastest Δ + 1 coloring algorithm running in time O(log Δ + √log n). The algorithm works without knowledge of χ and uses less than Δ colors, i.e., (1 - 1 O(χ))Δ with high probability. To the best of our knowledge this is the first distributed algorithm for (such) general graphs taking the chromatic number χ into account.",
"The following problem is considered: given a linked list of length n , compute the distance from each element of the linked list to the end of the list. The problem has two standard deterministic algorithms: a linear time serial algorithm, and an O (log n ) time parallel algorithm using n processors. We present new deterministic parallel algorithms for the problem. Our strongest results are (1) O (log n log* n ) time using n (log n log* n ) processors (this algorithm achieves optimal speed-up); (2) O (log n ) time using n log ( k ) n log n processors, for any fixed positive integer k . The algorithms apply a novel “random-like” deterministic technique. This technique provides for a fast and efficient breaking of an apparently symmetric situation in parallel and distributed computation.",
"We consider the distributed message-passing @math model. In this model a communication network is represented by a graph where vertices host processors, and communication is performed over the edges. Computation proceeds in synchronous rounds. The running time of an algorithm is the number of rounds from the beginning until all vertices terminate. Local computation is free. An algorithm is called local if it terminates within a constant number of rounds. The question of what problems can be computed locally was raised by Naor and Stockmeyer [16] in their seminal paper in STOC'93. Since then the quest for problems with local algorithms, and for problems that cannot be computed locally, has become a central research direction in the field of distributed algorithms [9,11,13,17]. We devise the first local algorithm for an NP-complete problem. Specifically, our randomized algorithm computes, with high probability, an O(n1 2+e ·χ)-coloring within O(1) rounds, where e>0 is an arbitrarily small constant, and χ is the chromatic number of the input graph. (This problem was shown to be NP-complete in [21].) On our way to this result we devise a constant-time algorithm for computing (O(1), O(n1 2+e))-network-decompositions. Network-decompositions were introduced by [1], and are very useful for solving various distributed problems. The best previously-known algorithm for network-decomposition has a polylogarithmic running time (but is applicable for a wider range of parameters) [15]. We also devise a Δ1+e-coloring algorithm for graphs with sufficiently large maximum degree Δ that runs within O(1) rounds. It improves the best previously-known result for this family of graphs, which is O(log*n) [19].",
"",
"This paper considers the computational power of anonymous message passing algorithms (henceforth, anonymous algorithms), i.e., distributed algorithms operating in a network of unidentified nodes. We prove that every problem that can be solved (and verified) by a randomized anonymous algorithm can also be solved by a deterministic anonymous algorithm provided that the latter is equipped with a 2-hop coloring of the input graph. Since the problem of 2-hop coloring a given graph (i.e., ensuring that two nodes with distance at most 2 have different colors) can by itself be solved by a randomized anonymous algorithm, it follows that with the exception of a few mock cases, the execution of every randomized anonymous algorithm can be decoupled into a generic preprocessing randomized stage that computes a 2-hop coloring, followed by a problem-specific deterministic stage. The main ingredient of our proof is a novel simulation method that relies on some surprising connections between 2-hop colorings and an extensively used graph lifting technique.",
"This paper concerns a number of algorithmic problems on graphs and how they may be solved in a distributed fashion. The computational model is such that each node of the graph is occupied by a processor which has its own ID. Processors are restricted to collecting data from others which are at a distance at most t away from them in t time units, but are otherwise computationally unbounded. This model focuses on the issue of locality in distributed processing, namely, to what extent a global solution to a computational problem can be obtained from locally available data.Three results are proved within this model: • A 3-coloring of an n-cycle requires time @math . This bound is tight, by previous work of Cole and Vishkin. • Any algorithm for coloring the d-regular tree of radius r which runs for time at most @math requires at least @math colors. • In an n-vertex graph of largest degree @math , an @math -coloring may be found in time @math .",
"The distributed @math -coloring problem is one of the most fundamental and well-studied problems in distributed algorithms. Starting with the work of Cole and Vishkin in 1986, a long line of gradually improving algorithms has been published. The state-of-the-art running time, prior to our work, is @math , due to Kuhn and Wattenhofer [Proceedings of the @math th Annual ACM Symposium on Principles of Distributed Computing, Denver, CO, 2006, pp. 7--15]. Linial [Proceedings of the @math th Annual IEEE Symposium on Foundation of Computer Science, Los Angeles, CA, 1987, pp. 331--335] proved a lower bound of @math for the problem, and Szegedy and Vishwanathan [Proceedings of the 25th Annual ACM Symposium on Theory of Computing, San Diego, CA, 1993, pp. 201--207] provided a heuristic argument that shows that algorithms from a wide family of locally iterative algorithms are unlikely to achieve a running time smaller than @math . We present a de...",
"",
"Consider an n-vertex graph G = (V, E) of maximum degree Δ, and suppose that each vertex v ∈ V hosts a processor. The processors are allowed to communicate only with their neighbors in G. The communication is synchronous, that is, it proceeds in discrete rounds. In the distributed vertex coloring problem, the objective is to color G with Δ + 1, or slightly more than Δ + 1, colors using as few rounds of communication as possible. (The number of rounds of communication will be henceforth referred to as running time.) Efficient randomized algorithms for this problem are known for more than twenty years [ 1986; Luby 1986]. Specifically, these algorithms produce a (Δ + 1)-coloring within O(log n) time, with high probability. On the other hand, the best known deterministic algorithm that requires polylogarithmic time employs O(Δ2) colors. This algorithm was devised in a seminal FOCS’87 paper by Linial [1987]. Its running time is O(log* n). In the same article, Linial asked whether one can color with significantly less than Δ2 colors in deterministic polylogarithmic time. By now, this question of Linial became one of the most central long-standing open questions in this area. In this article, we answer this question in the affirmative, and devise a deterministic algorithm that employs Δ1+o(1) colors, and runs in polylogarithmic time. Specifically, the running time of our algorithm is O(f(Δ)log Δ log n), for an arbitrarily slow-growing function f(Δ) = ω(1). We can also produce an O(Δ1+η)-coloring in O(log Δ log n)-time, for an arbitrarily small constant η > 0, and an O(Δ)-coloring in O(Δe log n) time, for an arbitrarily small constant e > 0. Our results are, in fact, far more general than this. In particular, for a graph of arboricity a, our algorithm produces an O(a1+η)-coloring, for an arbitrarily small constant η > 0, in time O(log a log n).",
"We give a new randomized distributed algorithm for (Δ +1)-coloring in the LOCAL model, running in O(s log Δ)+ 2O(slog log n) rounds in a graph of maximum degree Δ. This implies that the (Δ +1)-coloring problem is easier than the maximal independent set problem and the maximal matching problem, due to their lower bounds of Ω(min(sflog n log log n, flog Δ log log Δ)) by Kuhn, Moscibroda, and Wattenhofer [PODC’04]. Our algorithm also extends to list-coloring where the palette of each node contains Δ +1 colors. We extend the set of distributed symmetry-breaking techniques by performing a decomposition of graphs into dense and sparse parts.",
"Vertex coloring is a central concept in graph theory and an important symmetry-breaking primitive in distributed computing. Whereas degree-Δ graphs may require palettes of Δ + 1 colors in the worst case, it is well known that the chromatic number of many natural graph classes can be much smaller. In this paper we give new distributed algorithms to find ( Δ k ) -coloring in graphs of girth 4 (triangle-free graphs), girth 5, and trees. The parameter k can be at most ( 1 4 - o ( 1 ) ) ln ? Δ in triangle-free graphs and at most ( 1 - o ( 1 ) ) ln ? Δ in girth-5 graphs and trees, where o ( 1 ) is a function of Δ. Specifically, for Δ sufficiently large we can find such a coloring in O ( k + log * ? n ) time. Moreover, for any Δ we can compute such colorings in roughly logarithmic time for triangle-free and girth-5 graphs, and in O ( log ? Δ + log Δ ? log ? n ) time on trees. As a byproduct, our algorithm shows that the chromatic number of triangle-free graphs is at most ( 4 + o ( 1 ) ) Δ ln ? Δ , which improves on Jamall's recent bound of ( 67 + o ( 1 ) ) Δ ln ? Δ . Finally, we show that ( Δ + 1 ) -coloring for triangle-free graphs can be obtained in sublogarithmic time for any Δ.",
"Over the past 30 years numerous algorithms have been designed for symmetry breaking problems in the LOCAL model, such as maximal matching, MIS, vertex coloring, and edge-coloring. For most problems the best randomized algorithm is at least exponentially faster than the best deterministic algorithm. In this paper we prove that these exponential gaps are necessary and establish connections between the deterministic and randomized complexities in the LOCAL model. Each result has a very compelling take-away message: 1. Fast @math -coloring of trees requires random bits: Building on the recent lower bounds of , we prove that the randomized complexity of @math -coloring a tree with maximum degree @math is @math , whereas its deterministic complexity is @math for any @math . This also establishes a large separation between the deterministic complexity of @math -coloring and @math -coloring trees. 2. Randomized lower bounds imply deterministic lower bounds: We prove that any deterministic algorithm for a natural class of problems that runs in @math rounds can be transformed to run in @math rounds. If the transformed algorithm violates a lower bound (even allowing randomization), then one can conclude that the problem requires @math time deterministically. 3. Deterministic lower bounds imply randomized lower bounds: We prove that the randomized complexity of any natural problem on instances of size @math is at least its deterministic complexity on instances of size @math . This shows that a deterministic @math lower bound for any problem implies a randomized @math lower bound. It also illustrates that the graph shattering technique is absolutely essential to the LOCAL model.",
"Symmetry-breaking problems are among the most well studied in the field of distributed computing and yet the most fundamental questions about their complexity remain open. In this article we work in the LOCAL model (where the input graph and underlying distributed network are identical) and study the randomized complexity of four fundamental symmetry-breaking problems on graphs: computing MISs (maximal independent sets), maximal matchings, vertex colorings, and ruling sets. A small sample of our results includes the following: —An MIS algorithm running in O(log2D Δ L 2√log n, and comes close to the Ω(flog Δ log log Δ lower bound of Kuhn, Moscibroda, and Wattenhofer. —A maximal matching algorithm running in O(log Δ + log 4log n) time. This is the first significant improvement to the 1986 algorithm of Israeli and Itai. Moreover, its dependence on Δ is nearly optimal. —A (Δ + 1)-coloring algorithm requiring O(log Δ + 2o(√log log n) time, improving on an O(log Δ + √log n)-time algorithm of Schneider and Wattenhofer. —A method for reducing symmetry-breaking problems in low arboricity degeneracy graphs to low-degree graphs. (Roughly speaking, the arboricity or degeneracy of a graph bounds the density of any subgraph.) Corollaries of this reduction include an O(√log n)-time maximal matching algorithm for graphs with arboricity up to 2√log n and an O(log 2 3n)-time MIS algorithm for graphs with arboricity up to 2(log n)1 3. Each of our algorithms is based on a simple but powerful technique for reducing a randomized symmetry-breaking task to a corresponding deterministic one on a poly(log n)-size graph.",
"Numerous problems in Theoretical Computer Science can be solved very efficiently using powerful algebraic constructions. Computing shortest paths, constructing expanders, and proving the PCP Theorem, are just few examples of this phenomenon. The quest for combinatorial algorithms that do not use heavy algebraic machinery, but are roughly as efficient, has become a central field of study in this area. Combinatorial algorithms are often simpler than their algebraic counterparts. Moreover, in many cases, combinatorial algorithms and proofs provide additional understanding of studied problems. In this paper we initiate the study of combinatorial algorithms for Distributed Graph Coloring problems. In a distributed setting a communication network is modeled by a graph (G=(V,E) ) of maximum degree ( ). The vertices of (G ) host the processors, and communication is performed over the edges of (G ). The goal of distributed vertex coloring is to color (V ) with (( + 1) ) colors such that any two neighbors are colored with distinct colors. Currently, efficient algorithms for vertex coloring that require (O( + ^* n) ) time are based on the algebraic algorithm of Linial (SIAM J Comput 21(1):193–201, 1992) that employs set-systems. The best currently-known combinatorial set-system free algorithm, due to (SIAM J Discret Math 1(4):434–446, 1988), requires (O( ^2+ ^*n) ) time. We significantly improve over this by devising a combinatorial (( + 1) )-coloring algorithm that runs in (O( + ^* n) ) time. This exactly matches the running time of the best-known algebraic algorithm. In addition, we devise a tradeoff for computing (O( t) )-coloring in (O( t + ^* n) ) time, for almost the entire range (1 < t < ). We also compute a Maximal Independent Set in (O( + ^* n) ) time on general graphs, and in (O( n n) ) time on graphs of bounded arboricity. Prior to our work, these results could be only achieved using algebraic techniques. We believe that our algorithms are more suitable for real-life networks with limited resources, such as sensor networks."
]
} |
1705.05646 | 2616359666 | We present the first super-linear lower bounds for natural graph problems in the CONGEST model, answering a long-standing open question. Specifically, we show that any exact computation of a minimum vertex cover or a maximum independent set requires @math rounds in the worst case in the CONGEST model, as well as any algorithm for @math -coloring a graph, where @math is the chromatic number of the graph. We further show that such strong lower bounds are not limited to NP-hard problems, by showing two simple graph problems in P which require a quadratic and near-quadratic number of rounds. Finally, we address the problem of computing an exact solution to weighted all-pairs-shortest-paths (APSP), which arguably may be considered as a candidate for having a super-linear lower bound. We show a simple @math lower bound for this problem, which implies a separation between the weighted and unweighted cases, since the latter is known to have a complexity of @math . We also formally prove that the standard Alice-Bob framework is incapable of providing a super-linear lower bound for exact weighted APSP, whose complexity remains an intriguing open question. | Another classical problem in graph theory is finding a minimum vertex cover (MVC). In distributed computing, the time complexity of approximating MVC has been addressed in several cornerstone studies @cite_20 @cite_5 @cite_1 @cite_2 @cite_3 @cite_9 @cite_50 @cite_13 @cite_65 @cite_25 @cite_38 @cite_45 @cite_17 . | {
"cite_N": [
"@cite_38",
"@cite_9",
"@cite_1",
"@cite_65",
"@cite_3",
"@cite_45",
"@cite_50",
"@cite_2",
"@cite_5",
"@cite_13",
"@cite_25",
"@cite_20",
"@cite_17"
],
"mid": [
"2011595776",
"2029483371",
"2116636970",
"2101781054",
"",
"2006040238",
"1965897365",
"2055000959",
"",
"1957963525",
"2467514673",
"1813487646",
""
],
"abstract": [
"We show that maximal matchings can be computed deterministically in O(log4 n) rounds in the synchronous, message-passing model of computation. This is one of the very few cases known of a nontrivial graph structure, and the only \"classical\" one, which can be computed distributively in polylogarithmic time without recourse to randomization.",
"We give an efficient deterministic parallel approximation algorithm for the minimum-weight vertex- and set-cover problems and their duals (edge element packing). The algorithm is simple and suitable for distributed implementation. It fits no existing paradigm for fast, efficient parallel algorithms-it uses only \"local\" information at each step, yet is deterministic. (Generally, such algorithms have required randomization.) The result demonstrates that linear-programming primal-dual approximation techniques can lead to fast, efficient parallel algorithms. The presentation does not assume knowledge of such techniques.",
"We present a distributed algorithm that finds a maximal edge packing in O(Δ + log* W) synchronous communication rounds in a weighted graph, independent of the number of nodes in the network; here Δ is the maximum degree of the graph and W is the maximum weight. As a direct application, we have a distributed 2-approximation algorithm for minimum-weight vertex cover, with the same running time. We also show how to find an @math -approximation of minimum-weight set cover in O(f2k2 + fk log* W) rounds; here k is the maximum size of a subset in the set cover instance, f is the maximum frequency of an element, and W is the maximum weight of a subset. The algorithms are deterministic, and they can be applied in anonymous networks.",
"We present a local algorithm (constant-time distributed algorithm) for finding a 3-approximate vertex cover in bounded-degree graphs. The algorithm is deterministic, and no auxiliary information besides port numbering is required.",
"",
"We give simple, deterministic, distributed algorithms for computing maximal matchings, maximal independent sets and colourings. We show that edge colourings with at most 2Δ-1 colours, and maximal matchings can be computed within O(log* n + Δ) deterministic rounds, where Δ is the maximum degree of the network. We also show how to find maximal independent sets and (Δ + 1)-vertex colourings within O(log* n + Δ2) deterministic rounds. All hidden constants are very small and the algorithms are very simple.",
"The paper presents distributed and parallel δ-approximation algorithms for covering problems, where δ is the maximum number of variables on which any constraint depends (for example, δ = 2 for VERTEX COVER). Specific results include the following. ≺ For WEIGHTED VERTEX COVER, the first distributed 2-approximation algorithm taking O(log n) rounds and the first parallel 2-approximation algorithm in RNC. The algorithms generalize to covering mixed integer linear programs (CMIP) with two variables per constraint (δ = 2). ≺ For any covering problem with monotone constraints and submodular cost, a distributed δ-approximation algorithm taking O(log2 |C|) rounds, where |C| is the number of constraints. (Special cases include CMIP, facility location, and probabilistic (two-stage) variants of these problems.)",
"In this article, we consider the problem of computing a minimum-weight vertex-cover in an n-node, weighted, undirected graph G e (V,E). We present a fully distributed algorithm for computing vertex covers of weight at most twice the optimum, in the case of integer weights. Our algorithm runs in an expected number of O(log n + log W) communication rounds, where W is the average vertex-weight. The previous best algorithm for this problem requires O(log n(log n + logW)) rounds and it is not fully distributed. For a maximal matching M in G, it is a well-known fact that any vertex-cover in G needs to have at least vMv vertices. Our algorithm is based on a generalization of this combinatorial lower-bound to the weighted setting.",
"",
"The question of what can be computed, and how efficiently, is at the core of computer science. Not surprisingly, in distributed systems and networking research, an equally fundamental question is what can be computed in a distributed fashion. More precisely, if nodes of a network must base their decision on information in their local neighborhood only, how well can they compute or approximate a global (optimization) problem? In this paper we give the first polylogarithmic lower bound on such local computation for (optimization) problems including minimum vertex cover, minimum (connected) dominating set, maximum matching, maximal independent set, and maximal matching. In addition, we present a new distributed algorithm for solving general covering and packing linear programs. For some problems this algorithm is tight with the lower bounds, whereas for others it is a distributed approximation scheme. Together, our lower and upper bounds establish the local computability and approximability of a large class of problems, characterizing how much local information is required to solve these tasks.",
"Symmetry-breaking problems are among the most well studied in the field of distributed computing and yet the most fundamental questions about their complexity remain open. In this article we work in the LOCAL model (where the input graph and underlying distributed network are identical) and study the randomized complexity of four fundamental symmetry-breaking problems on graphs: computing MISs (maximal independent sets), maximal matchings, vertex colorings, and ruling sets. A small sample of our results includes the following: —An MIS algorithm running in O(log2D Δ L 2√log n, and comes close to the Ω(flog Δ log log Δ lower bound of Kuhn, Moscibroda, and Wattenhofer. —A maximal matching algorithm running in O(log Δ + log 4log n) time. This is the first significant improvement to the 1986 algorithm of Israeli and Itai. Moreover, its dependence on Δ is nearly optimal. —A (Δ + 1)-coloring algorithm requiring O(log Δ + 2o(√log log n) time, improving on an O(log Δ + √log n)-time algorithm of Schneider and Wattenhofer. —A method for reducing symmetry-breaking problems in low arboricity degeneracy graphs to low-degree graphs. (Roughly speaking, the arboricity or degeneracy of a graph bounds the density of any subgraph.) Corollaries of this reduction include an O(√log n)-time maximal matching algorithm for graphs with arboricity up to 2√log n and an O(log 2 3n)-time MIS algorithm for graphs with arboricity up to 2(log n)1 3. Each of our algorithms is based on a simple but powerful technique for reducing a randomized symmetry-breaking task to a corresponding deterministic one on a poly(log n)-size graph.",
"We present a distributed 2-approximation algorithm for the minimum vertex cover problem. The algorithm is deterministic, and it runs in (Δ + 1)2 synchronous communication rounds, where Δ is the maximum degree of the graph. For Δ = 3, we give a 2-approximation algorithm also for the weighted version of the problem.",
""
]
} |
1705.05646 | 2616359666 | We present the first super-linear lower bounds for natural graph problems in the CONGEST model, answering a long-standing open question. Specifically, we show that any exact computation of a minimum vertex cover or a maximum independent set requires @math rounds in the worst case in the CONGEST model, as well as any algorithm for @math -coloring a graph, where @math is the chromatic number of the graph. We further show that such strong lower bounds are not limited to NP-hard problems, by showing two simple graph problems in P which require a quadratic and near-quadratic number of rounds. Finally, we address the problem of computing an exact solution to weighted all-pairs-shortest-paths (APSP), which arguably may be considered as a candidate for having a super-linear lower bound. We show a simple @math lower bound for this problem, which implies a separation between the weighted and unweighted cases, since the latter is known to have a complexity of @math . We also formally prove that the standard Alice-Bob framework is incapable of providing a super-linear lower bound for exact weighted APSP, whose complexity remains an intriguing open question. | Observe that finding a minimum size vertex cover is equivalent to finding a maximum size independent set. However, these problems are not equivalent in an approximation-preserving way. Distributed approximations for maximum independent set has been studied in @cite_42 @cite_43 @cite_53 @cite_0 . | {
"cite_N": [
"@cite_0",
"@cite_43",
"@cite_53",
"@cite_42"
],
"mid": [
"2952428427",
"1869515244",
"2503706701",
"1502920553"
],
"abstract": [
"We present a simple distributed @math -approximation algorithm for maximum weight independent set (MaxIS) in the @math model which completes in @math rounds, where @math is the maximum degree, @math is the number of rounds needed to compute a maximal independent set (MIS) on @math , and @math is the maximum weight of a node. Whether our algorithm is randomized or deterministic depends on the MIS algorithm used as a black-box. Plugging in the best known algorithm for MIS gives a randomized solution in @math rounds, where @math is the number of nodes. We also present a deterministic @math -round algorithm based on coloring. We then show how to use our MaxIS approximation algorithms to compute a @math -approximation for maximum weight matching without incurring any additional round penalty in the @math model. We use a known reduction for simulating algorithms on the line graph while incurring congestion, but we show our algorithm is part of a broad family of for which we describe a mechanism that allows the simulation to run in the @math model without an additional overhead. Next, we show that for maximum weight matching, relaxing the approximation factor to ( @math ) allows us to devise a distributed algorithm requiring @math rounds for any constant @math . For the unweighted case, we can even obtain a @math -approximation in this number of rounds. These algorithms are the first to achieve the provably optimal round complexity with respect to dependency on @math .",
"We give deterministic distributed algorithms that given i¾?> 0 find in a planar graph G, (1±i¾?)-approximations of a maximum independent set, a maximum matching, and a minimum dominating set. The algorithms run in O(log*|G|) rounds. In addition, we prove that no faster deterministic approximation is possible and show that if randomization is allowed it is possible to beat the lower bound for deterministic algorithms.",
"We show that the first phase of the Linial-Saks network decomposition algorithm gives a randomized distributed O(ne)-approximation algorithm for the maximum independent set problem that operates in O(1 e) rounds, and we give a matching lower bound that holds even for bipartite graphs.",
"In this paper we extend the lower bound technique by Linial for local coloring and maximal independent sets. We show that constant approximations to maximum independent sets on a ring require at least log-star time. More generally, the product of approximation quality and running time cannot be less than log-star. Using a generalized ring topology, we gain identical lower bounds for approximations to minimum dominating sets. Since our generalized ring topology is contained in a number of geometric graphs such as the unit disk graph, our bounds directly apply as lower bounds for quite a few algorithmic problems in wireless networking. Having in mind these and other results about local approximations of maximum independent sets and minimum dominating sets, one might think that the former are always at least as difficult to obtain as the latter. Conversely, we show that graphs exist, where a maximum independent set can be determined without any communication, while finding even an approximation to a minimum dominating set is as hard as in general graphs."
]
} |
1705.05646 | 2616359666 | We present the first super-linear lower bounds for natural graph problems in the CONGEST model, answering a long-standing open question. Specifically, we show that any exact computation of a minimum vertex cover or a maximum independent set requires @math rounds in the worst case in the CONGEST model, as well as any algorithm for @math -coloring a graph, where @math is the chromatic number of the graph. We further show that such strong lower bounds are not limited to NP-hard problems, by showing two simple graph problems in P which require a quadratic and near-quadratic number of rounds. Finally, we address the problem of computing an exact solution to weighted all-pairs-shortest-paths (APSP), which arguably may be considered as a candidate for having a super-linear lower bound. We show a simple @math lower bound for this problem, which implies a separation between the weighted and unweighted cases, since the latter is known to have a complexity of @math . We also formally prove that the standard Alice-Bob framework is incapable of providing a super-linear lower bound for exact weighted APSP, whose complexity remains an intriguing open question. | Distance computation problems have been widely studied in the CONGEST model for both weighted and unweighted networks @cite_49 @cite_32 @cite_59 @cite_63 @cite_37 @cite_4 @cite_41 @cite_60 @cite_24 @cite_34 @cite_58 . One of the most fundamental problems of distance computations is computing all pairs shortest paths. For unweighted networks, an upper bound of @math was recently shown by @cite_34 , matching the lower bound of @cite_32 . Moreover, the possibility of bypassing this near-linear barrier for any constant approximation factor was ruled out by @cite_24 . For the weighted case, however, we are still very far from understanding the complexity of APSP, as there is still a huge gap between the upper and lower bounds. Recently, Elkin @cite_47 showed an @math upper bound for weighted APSP, while the previously highest lower bound was the near-linear lower bound of @cite_24 (which holds also for any @math -approximation factor in the weighted case). | {
"cite_N": [
"@cite_37",
"@cite_4",
"@cite_60",
"@cite_41",
"@cite_58",
"@cite_32",
"@cite_24",
"@cite_63",
"@cite_59",
"@cite_49",
"@cite_47",
"@cite_34"
],
"mid": [
"",
"1515635142",
"2107282727",
"2949588463",
"",
"1582638066",
"2040011014",
"2397361776",
"2048098617",
"2399682168",
"2952103668",
"2474966478"
],
"abstract": [
"",
"We study the broadcast version of the CONGEST CLIQUE model of distributed computing. In this model, in each round, any node in a network of size @math can send the same message (i.e. broadcast a message) of limited size to every other node in the network. Nanongkai presented in [STOC'14] a randomized @math -approximation algorithm to compute all pairs shortest paths (APSP) in time @math on weighted graphs, where we use the convention that @math is essentially @math polylog @math and @math is essentially @math polylog @math . We complement this result by proving that any randomized @math -approximation of APSP and @math -approximation of the diameter of a graph takes @math time in the worst case. This demonstrates that getting a negligible improvement in the approximation factor requires significantly more time. Furthermore this bound implies that already computing a @math -approximation of all pairs shortest paths is among the hardest graph-problems in the broadcast-version of the CONGEST CLIQUE model and contrasts a recent @math -approximation for APSP that runs in time @math in the unicast version of the CONGEST CLIQUE model. On the positive side we provide a deterministic version of Nanongkai's @math -approximation algorithm for APSP. To do so we present a fast deterministic construction of small hitting sets. We also show how to replace another randomized part within Nanongkai's algorithm with a deterministic source-detection algorithm designed for the CONGEST model presented by Lenzen and Peleg at PODC'13.",
"Given a simple graph G=(V,E) and a set of sources S ⊆ V, denote for each node ν e V by Lν(∞) the lexicographically ordered list of distance source pairs (d(s,v),s), where s ∈ S. For integers d,k ∈ N∪ ∞ , we consider the source detection, or (S,d,k)-detection task, requiring each node v to learn the first k entries of Lν(∞) (if for all of them d(s,v) ≤ d) or all entries (d(s,v),s) ∈ Lν(∞) satisfying that d(s,v) ≤ d (otherwise). Solutions to this problem provide natural generalizations of concurrent breadth-first search (BFS) tree constructions. For example, the special case of k=∞ requires each source s ∈ S to build a complete BFS tree rooted at s, whereas the special case of d=∞ and S=V requires constructing a partial BFS tree comprising at least k nodes from every node in V. In this work, we give a simple, near-optimal solution for the source detection task in the CONGEST model, where messages contain at most O(log n) bits, running in d+k rounds. We demonstrate its utility for various routing problems, exact and approximate diameter computation, and spanner construction. For those problems, we obtain algorithms in the CONGEST model that are faster and in some cases much simpler than previous solutions.",
"We study approximate distributed solutions to the weighted all-pairs-shortest-paths (APSP) problem in the CONGEST model. We obtain the following results. @math A deterministic @math -approximation to APSP in @math rounds. This improves over the best previously known algorithm, by both derandomizing it and by reducing the running time by a @math factor. In many cases, routing schemes involve relabeling, i.e., assigning new names to nodes and require that these names are used in distance and routing queries. It is known that relabeling is necessary to achieve running times of @math . In the relabeling model, we obtain the following results. @math A randomized @math -approximation to APSP, for any integer @math , running in @math rounds, where @math is the hop diameter of the network. This algorithm simplifies the best previously known result and reduces its approximation ratio from @math to @math . Also, the new algorithm uses uses labels of asymptotically optimal size, namely @math bits. @math A randomized @math -approximation to APSP, for any integer @math , running in time @math and producing compact routing tables of size @math . The node lables consist of @math bits. This improves on the approximation ratio of @math for tables of that size achieved by the best previously known algorithm, which terminates faster, in @math rounds.",
"",
"We study the problem of computing the diameter of a network in a distributed way. The model of distributed computation we consider is: in each synchronous round, each node can transmit a different (but short) message to each of its neighbors. We provide an Ω(n) lower bound for the number of communication rounds needed, where n denotes the number of nodes in the network. This lower bound is valid even if the diameter of the network is a small constant. We also show that a (3 2 − e)-approximation of the diameter requires Ω (√n + D) rounds. Furthermore we use our new technique to prove an Ω (√n + D) lower bound on approximating the girth of a graph by a factor 2 − e.",
"A distributed network is modeled by a graph having n nodes (processors) and diameter D. We study the time complexity of approximating weighted (undirected) shortest paths on distributed networks with a O (log n) bandwidth restriction on edges (the standard synchronous CONGEST model). The question whether approximation algorithms help speed up the shortest paths and distance computation (more precisely distance computation) was raised since at least 2004 by Elkin (SIGACT News 2004). The unweighted case of this problem is well-understood while its weighted counterpart is fundamental problem in the area of distributed approximation algorithms and remains widely open. We present new algorithms for computing both single-source shortest paths (SSSP) and all-pairs shortest paths (APSP) in the weighted case. Our main result is an algorithm for SSSP. Previous results are the classic O(n)-time Bellman-Ford algorithm and an O(n1 2+1 2k + D)-time (8k⌈log(k + 1)⌉ --1)-approximation algorithm, for any integer k ≥ 1, which follows from the result of Lenzen and Patt-Shamir (STOC 2013). (Note that Lenzen and Patt-Shamir in fact solve a harder problem, and we use O(·) to hide the O(poly log n) term.) We present an O (n1 2D1 4 + D)-time (1 + o(1))-approximation algorithm for SSSP. This algorithm is sublinear-time as long as D is sublinear, thus yielding a sublinear-time algorithm with almost optimal solution. When D is small, our running time matches the lower bound of Ω(n1 2 + D) by Das (SICOMP 2012), which holds even when D=Θ(log n), up to a poly log n factor. As a by-product of our technique, we obtain a simple O (n)-time (1+ o(1))-approximation algorithm for APSP, improving the previous O(n)-time O(1)-approximation algorithm following from the results of Lenzen and Patt-Shamir. We also prove a matching lower bound. Our techniques also yield an O(n1 2) time algorithm on fully-connected networks, which guarantees an exact solution for SSSP and a (2+ o(1))-approximate solution for APSP. All our algorithms rely on two new simple tools: light-weight algorithm for bounded-hop SSSP and shortest-path diameter reduction via shortcuts. These tools might be of an independent interest and useful in designing other distributed algorithms.",
"",
"We present an algorithm to compute All Pairs Shortest Paths (APSP) of a network in a distributed way. The model of distributed computation we consider is the message passing model: in each synchronous round, every node can transmit a different (but short) message to each of its neighbors. We provide an algorithm that computes APSP in O(n) communication rounds, where n denotes the number of nodes in the network. This implies a linear time algorithm for computing the diameter of a network. Due to a lower bound these two algorithms are optimal up to a logarithmic factor. Furthermore, we present a new lower bound for approximating the diameter D of a graph: Being allowed to answer D+1 or D can speed up the computation by at most a factor D. On the positive side, we provide an algorithm that achieves such a speedup of D and computes an (1+eepsilon) multiplicative approximation of the diameter. We extend these algorithms to compute or approximate other problems, such as girth, radius, center and peripheral vertices. At the heart of these approximation algorithms is the S-Shortest Paths problem which we solve in O(|S|+D) time.",
"We develop a new technique for constructing sparse graphs that allow us to prove near-linear lower bounds on the round complexity of computing distances in the CONGEST model. Specifically, we show an @math lower bound for computing the diameter in sparse networks, which was previously known only for dense networks [, SODA 2012]. In fact, we can even modify our construction to obtain graphs with constant degree, using a simple but powerful degree-reduction technique which we define. Moreover, our technique allows us to show @math lower bounds for computing @math -approximations of the diameter or the radius, and for computing a @math -approximation of all eccentricities. For radius, we are unaware of any previous lower bounds. For diameter, these greatly improve upon previous lower bounds and are tight up to polylogarithmic factors [, SODA 2012], and for eccentricities the improvement is both in the lower bound and in the approximation factor [Holzer and Wattenhofer, PODC 2012]. Interestingly, our technique also allows showing an almost-linear lower bound for the verification of @math -spanners, for @math .",
"The distributed single-source shortest paths problem is one of the most fundamental and central problems in the message-passing distributed computing. Classical Bellman-Ford algorithm solves it in @math time, where @math is the number of vertices in the input graph @math . Peleg and Rubinovich (FOCS'99) showed a lower bound of @math for this problem, where @math is the hop-diameter of @math . Whether or not this problem can be solved in @math time when @math is relatively small is a major notorious open question. Despite intensive research LP13,N14,HKN15,EN16,BKKL16 that yielded near-optimal algorithms for the approximate variant of this problem, no progress was reported for the original problem. In this paper we answer this question in the affirmative. We devise an algorithm that requires @math time, for @math , and @math time, for larger @math . This running time is sublinear in @math in almost the entire range of parameters, specifically, for @math . For the all-pairs shortest paths problem, our algorithm requires @math time, regardless of the value of @math . We also devise the first algorithm with non-trivial complexity guarantees for computing exact shortest paths in the multipass semi-streaming model of computation. From the technical viewpoint, our algorithm computes a hopset @math of a skeleton graph @math of @math without first computing @math itself. We then conduct a Bellman-Ford exploration in @math , while computing the required edges of @math on the fly. As a result, our algorithm computes exactly those edges of @math that it really needs, rather than computing approximately the entire @math .",
"Given an unweighted and undirected graph, this paper aims to give a tight distributed algorithm for computing the all pairs shortest paths (APSP) under synchronous communications and the CONGEST(B) model, where each node can only transfer B bits of information along each incident edge in a round. The best previous results for distributively computing APSP need O(N+D) time where N is the number of nodes and D is the diameter [1,2]. However, there is still a B factor gap from the lower bound Ω(N B+D) [1]. In order to close this gap, we propose a multiplexing technique to push the parallelization of distributed BFS tree constructions to the limit such that we can solve APSP in O(N B+D) time which meets the lower bound. This result also implies a Θ(N B+D) time distributed algorithm for diameter. In addition, we extend our distributed algorithm to compute girth which is the length of the shortest cycle and clustering coefficient (CC) which is related to counting the number of triangles incident to each node. The time complexities for computing these two graph properties are also O(N B+D)."
]
} |
1705.05646 | 2616359666 | We present the first super-linear lower bounds for natural graph problems in the CONGEST model, answering a long-standing open question. Specifically, we show that any exact computation of a minimum vertex cover or a maximum independent set requires @math rounds in the worst case in the CONGEST model, as well as any algorithm for @math -coloring a graph, where @math is the chromatic number of the graph. We further show that such strong lower bounds are not limited to NP-hard problems, by showing two simple graph problems in P which require a quadratic and near-quadratic number of rounds. Finally, we address the problem of computing an exact solution to weighted all-pairs-shortest-paths (APSP), which arguably may be considered as a candidate for having a super-linear lower bound. We show a simple @math lower bound for this problem, which implies a separation between the weighted and unweighted cases, since the latter is known to have a complexity of @math . We also formally prove that the standard Alice-Bob framework is incapable of providing a super-linear lower bound for exact weighted APSP, whose complexity remains an intriguing open question. | Distance computation problems have also been considered in the CONGESTED-CLIQUE model @cite_58 @cite_54 @cite_4 , in which the underlying communication network forms a clique. In this model @cite_54 showed that unweighted APSP, and a @math -approximation for weighted APSP, can be computed in @math rounds. | {
"cite_N": [
"@cite_54",
"@cite_58",
"@cite_4"
],
"mid": [
"2950813619",
"",
"1515635142"
],
"abstract": [
"In this work, we use algebraic methods for studying distance computation and subgraph detection tasks in the congested clique model. Specifically, we adapt parallel matrix multiplication implementations to the congested clique, obtaining an @math round matrix multiplication algorithm, where @math is the exponent of matrix multiplication. In conjunction with known techniques from centralised algorithmics, this gives significant improvements over previous best upper bounds in the congested clique model. The highlight results include: -- triangle and 4-cycle counting in @math rounds, improving upon the @math triangle detection algorithm of [DISC 2012], -- a @math -approximation of all-pairs shortest paths in @math rounds, improving upon the @math -round @math -approximation algorithm of Nanongkai [STOC 2014], and -- computing the girth in @math rounds, which is the first non-trivial solution in this model. In addition, we present a novel constant-round combinatorial algorithm for detecting 4-cycles.",
"",
"We study the broadcast version of the CONGEST CLIQUE model of distributed computing. In this model, in each round, any node in a network of size @math can send the same message (i.e. broadcast a message) of limited size to every other node in the network. Nanongkai presented in [STOC'14] a randomized @math -approximation algorithm to compute all pairs shortest paths (APSP) in time @math on weighted graphs, where we use the convention that @math is essentially @math polylog @math and @math is essentially @math polylog @math . We complement this result by proving that any randomized @math -approximation of APSP and @math -approximation of the diameter of a graph takes @math time in the worst case. This demonstrates that getting a negligible improvement in the approximation factor requires significantly more time. Furthermore this bound implies that already computing a @math -approximation of all pairs shortest paths is among the hardest graph-problems in the broadcast-version of the CONGEST CLIQUE model and contrasts a recent @math -approximation for APSP that runs in time @math in the unicast version of the CONGEST CLIQUE model. On the positive side we provide a deterministic version of Nanongkai's @math -approximation algorithm for APSP. To do so we present a fast deterministic construction of small hitting sets. We also show how to replace another randomized part within Nanongkai's algorithm with a deterministic source-detection algorithm designed for the CONGEST model presented by Lenzen and Peleg at PODC'13."
]
} |
1705.05646 | 2616359666 | We present the first super-linear lower bounds for natural graph problems in the CONGEST model, answering a long-standing open question. Specifically, we show that any exact computation of a minimum vertex cover or a maximum independent set requires @math rounds in the worst case in the CONGEST model, as well as any algorithm for @math -coloring a graph, where @math is the chromatic number of the graph. We further show that such strong lower bounds are not limited to NP-hard problems, by showing two simple graph problems in P which require a quadratic and near-quadratic number of rounds. Finally, we address the problem of computing an exact solution to weighted all-pairs-shortest-paths (APSP), which arguably may be considered as a candidate for having a super-linear lower bound. We show a simple @math lower bound for this problem, which implies a separation between the weighted and unweighted cases, since the latter is known to have a complexity of @math . We also formally prove that the standard Alice-Bob framework is incapable of providing a super-linear lower bound for exact weighted APSP, whose complexity remains an intriguing open question. | The problem of finding subgraphs of a certain topology has received a lot of attention in both the sequential and the distributed settings (see, e.g., @cite_55 @cite_30 @cite_56 @cite_18 @cite_12 @cite_28 @cite_21 @cite_19 @cite_15 @cite_54 and references therein). The problems of finding paths of length 4 or 5 with zero weight are also related to other fundamental problems, notable in our context is APSP @cite_55 . | {
"cite_N": [
"@cite_30",
"@cite_18",
"@cite_28",
"@cite_55",
"@cite_21",
"@cite_54",
"@cite_56",
"@cite_19",
"@cite_15",
"@cite_12"
],
"mid": [
"1996791172",
"2049661027",
"2949916381",
"2952976587",
"2107805020",
"2950813619",
"",
"2949944845",
"2015960500",
"1799262171"
],
"abstract": [
"For a pattern graph @math on @math nodes, we consider the problems of finding and counting the number of (not necessarily induced) copies of @math in a given large graph @math on @math nodes, as well as finding minimum weight copies in both node-weighted and edge-weighted graphs. Our results include the following: 1. The number of copies of an @math with an independent set of size @math can be computed exactly in @math time and @math space, or in @math time and @math space. (The @math notation omits @math factors.) To obtain these algorithms we provide fast algorithms for computing the permanent of a @math matrix over rings and semirings. 2. The number of copies of any @math having minimum (or maximum) node-weight (with arbitrary real weights on nodes) can be found in @math time, where @math is the matrix multiplication exponent and @math is divisible by @math . Similar results hold for other values of @math . Also, the number...",
"The 3SUM problem is to decide, given a set of @math real numbers, whether any three sum to zero. It is widely conjectured that a trivial @math -time algorithm is optimal and over the years the consequences of this conjecture have been revealed. This 3SUM conjecture implies @math lower bounds on numerous problems in computational geometry and a variant of the conjecture implies strong lower bounds on triangle enumeration, dynamic graph algorithms, and string matching data structures. In this paper we refute the 3SUM conjecture. We prove that the decision tree complexity of 3SUM is @math and give two subquadratic 3SUM algorithms, a deterministic one running in @math time and a randomized one running in @math time with high probability. Our results lead directly to improved bounds for @math -variate linear degeneracy testing for all odd @math . The problem is to decide, given a linear function @math and a set @math , whether @math . We show the decision tree complexity of this problem is @math . Finally, we give a subcubic algorithm for a generalization of the @math -product over real-valued matrices and apply it to the problem of finding zero-weight triangles in weighted graphs. We give a depth- @math decision tree for this problem, as well as an algorithm running in time @math .",
"We present a new way to encode weighted sums into unweighted pairwise constraints, obtaining the following results. - Define the k-SUM problem to be: given n integers in [-n^2k, n^2k] are there k which sum to zero? (It is well known that the same problem over arbitrary integers is equivalent to the above definition, by linear-time randomized reductions.) We prove that this definition of k-SUM remains W[1]-hard, and is in fact W[1]-complete: k-SUM can be reduced to f(k) * n^o(1) instances of k-Clique. - The maximum node-weighted k-Clique and node-weighted k-dominating set problems can be reduced to n^o(1) instances of the unweighted k-Clique and k-dominating set problems, respectively. This implies a strong equivalence between the time complexities of the node weighted problems and the unweighted problems: any polynomial improvement on one would imply an improvement for the other. - A triangle of weight 0 in a node weighted graph with m edges can be deterministically found in m^1.41 time.",
"We consider the Exact-Weight-H problem of finding a (not necessarily induced) subgraph H of weight 0 in an edge-weighted graph G. We show that for every H, the complexity of this problem is strongly related to that of the infamous k-Sum problem. In particular, we show that under the k-Sum Conjecture, we can achieve tight upper and lower bounds for the Exact-Weight-H problem for various subgraphs H such as matching, star, path, and cycle. One interesting consequence is that improving on the O(n^3) upper bound for Exact-Weight-4-Path or Exact-Weight-5-Path will imply improved algorithms for 3-Sum, 5-Sum, All-Pairs Shortest Paths and other fundamental problems. This is in sharp contrast to the minimum-weight and (unweighted) detection versions, which can be solved easily in time O(n^2). We also show that a faster algorithm for any of the following three problems would yield faster algorithms for the others: 3-Sum, Exact-Weight-3-Matching, and Exact-Weight-3-Star.",
"We study the computation power of the congested clique, a model of distributed computation where n players communicate with each other over a complete network in order to compute some function of their inputs. The number of bits that can be sent on any edge in a round is bounded by a parameter b We consider two versions of the model: in the first, the players communicate by unicast, allowing them to send a different message on each of their links in one round; in the second, the players communicate by broadcast, sending one message to all their neighbors. It is known that the unicast version of the model is quite powerful; to date, no lower bounds for this model are known. In this paper we provide a partial explanation by showing that the unicast congested clique can simulate powerful classes of bounded-depth circuits, implying that even slightly super-constant lower bounds for the congested clique would give new lower bounds in circuit complexity. Moreover, under a widely-believed conjecture on matrix multiplication, the triangle detection problem, studied in [8], can be solved in O(ne) time for any e > 0. The broadcast version of the congested clique is the well-known multi-party shared-blackboard model of communication complexity (with number-in-hand input). This version is more amenable to lower bounds, and in this paper we show that the subgraph detection problem studied in [8] requires polynomially many rounds for several classes of subgraphs. We also give upper bounds for the subgraph detection problem, and relate the hardness of triangle detection in the broadcast congested clique to the communication complexity of set disjointness in the 3-party number-on-forehead model.",
"In this work, we use algebraic methods for studying distance computation and subgraph detection tasks in the congested clique model. Specifically, we adapt parallel matrix multiplication implementations to the congested clique, obtaining an @math round matrix multiplication algorithm, where @math is the exponent of matrix multiplication. In conjunction with known techniques from centralised algorithmics, this gives significant improvements over previous best upper bounds in the congested clique model. The highlight results include: -- triangle and 4-cycle counting in @math rounds, improving upon the @math triangle detection algorithm of [DISC 2012], -- a @math -approximation of all-pairs shortest paths in @math rounds, improving upon the @math -round @math -approximation algorithm of Nanongkai [STOC 2014], and -- computing the girth in @math rounds, which is the first non-trivial solution in this model. In addition, we present a novel constant-round combinatorial algorithm for detecting 4-cycles.",
"",
"Let G = (V,E) be an n-vertex graph and M_d a d-vertex graph, for some constant d. Is M_d a subgraph of G? We consider this problem in a model where all n processes are connected to all other processes, and each message contains up to O(log n) bits. A simple deterministic algorithm that requires O(n^((d-2) d) log n) communication rounds is presented. For the special case that M_d is a triangle, we present a probabilistic algorithm that requires an expected O(ceil(n^(1 3) (t^(2 3) + 1))) rounds of communication, where t is the number of triangles in the graph, and O(min n^(1 3) log^(2 3) n (t^(2 3) + 1), n^(1 3) ) with high probability. We also present deterministic algorithms specially suited for sparse graphs. In any graph of maximum degree Delta, we can test for arbitrary subgraphs of diameter D in O(ceil(Delta^(D+1) n)) rounds. For triangles, we devise an algorithm featuring a round complexity of O(A^2 n + log_(2+n A^2) n), where A denotes the arboricity of G.",
"We present an assortment of methods for finding and counting simple cycles of a given length in directed and undirected graphs. Most of the bounds obtained depend solely on the number of edges in the graph in question, and not on the number of vertices. The bounds obtained improve upon various previously known results.",
"Given two graphs @math and @math , the Subgraph Isomorphism problem asks if @math is isomorphic to a subgraph of @math . While NP-hard in general, algorithms exist for various parameterized versions of the problem: for example, the problem can be solved (1) in time @math using the color-coding technique of Alon, Yuster, and Zwick; (2) in time @math using Courcelle's Theorem; (3) in time @math using a result on first-order model checking by Frick and Grohe; or (4) in time @math for connected @math using the algorithm of Matou s ek and Thomas. Already this small sample of results shows that the way an algorithm can depend on the parameters is highly nontrivial and subtle. We develop a framework involving 10 relevant parameters for each of @math and @math (such as treewidth, pathwidth, genus, maximum degree, number of vertices, number of components, etc.), and ask if an algorithm with running time [ f_1(p_1,p_2,..., p_ ) n^ f_2(p_ +1 ,..., p_k) ] exist, where each of @math is one of the 10 parameters depending only on @math or @math . We show that all the questions arising in this framework are answered by a set of 11 maximal positive results (algorithms) and a set of 17 maximal negative results (hardness proofs); some of these results already appear in the literature, while others are new in this paper. On the algorithmic side, our study reveals for example that an unexpected combination of bounded degree, genus, and feedback vertex set number of @math gives rise to a highly nontrivial algorithm for Subgraph Isomorphism. On the hardness side, we present W[1]-hardness proofs under extremely restricted conditions, such as when @math is a bounded-degree tree of constant pathwidth and @math is a planar graph of bounded pathwidth."
]
} |
1705.04927 | 2614901197 | In this paper, we present a new mathematical foundation for image-based lighting. Using a simple manipulation of the local coordinate system, we derive a closed-form solution to the light integral equation under distant environment illumination. We derive our solution for different BRDF's such as lambertian and Phong-like. The method is free of noise, and provides the possibility of using the full spectrum of frequencies captured by images taken from the environment. This allows for the color of the rendered object to be toned according to the color of the light in the environment. Experimental results also show that one can gain an order of magnitude or higher in rendering time compared to Monte Carlo quadrature methods and spherical harmonics. | Some very interesting closed-form solutions have already been proposed to solve the light integral. The appeal of closed-form solutions lies in the fact that they provide complete elimination of noise. Furthermore, the availability of a closed-form solution expedites the rendering process significantly. For estimation methods such as Monte Carlo @cite_112 , many samples of the environment are required to render realistic low-noise images. On the other hand, several hours might be required to generate one single image. | {
"cite_N": [
"@cite_112"
],
"mid": [
"2033102965"
],
"abstract": [
"In a distributed ray tracer, the sampling strategy is the crucial part of the direct lighting calculation. Monte Carlo integration with importance sampling is used to carry out this calculation. Importance sampling involves the design of integrand-specific probability density functions that are used to generate sample points for the numerical quadrature. Probability density functions are presented that aid in the direct lighting calculation from luminaires of various simple shapes. A method for defining a probability density function over a set of luminaires is presented that allows the direct lighting calculation to be carried out with a number of sample points that is independent of the number of luminaires."
]
} |
1705.04927 | 2614901197 | In this paper, we present a new mathematical foundation for image-based lighting. Using a simple manipulation of the local coordinate system, we derive a closed-form solution to the light integral equation under distant environment illumination. We derive our solution for different BRDF's such as lambertian and Phong-like. The method is free of noise, and provides the possibility of using the full spectrum of frequencies captured by images taken from the environment. This allows for the color of the rendered object to be toned according to the color of the light in the environment. Experimental results also show that one can gain an order of magnitude or higher in rendering time compared to Monte Carlo quadrature methods and spherical harmonics. | Currently, the closed-form solutions proposed in the literature mostly target specific scenarios. For example, the work done in @cite_5 targets linear light sources and provides a solution to the integral for diffuse and specular materials lit by such a light. The work done by Arvo @cite_114 provides analytic solutions to the light integral for polyhedral sources using the irradiance jacobian. @cite_62 introduced the use of B-splines to represent surface radiance in static scenes. A recent work by @cite_137 provides a closed-form solution to the light integral given isotropic point light sources. Their solution targets the scenarios of fog, mist and haze. | {
"cite_N": [
"@cite_137",
"@cite_5",
"@cite_62",
"@cite_114"
],
"mid": [
"2170961588",
"2002071611",
"199205947",
"2170485458"
],
"abstract": [
"We consider real-time rendering of scenes in participating media, capturing the effects of light scattering in fog, mist and haze. While a number of sophisticated approaches based on Monte Carlo and finite element simulation have been developed, those methods do not work at interactive rates. The most common real-time methods are essentially simple variants of the OpenGL fog model. While easy to use and specify, that model excludes many important qualitative effects like glows around light sources, the impact of volumetric scattering on the appearance of surfaces such as the diffusing of glossy highlights, and the appearance under complex lighting such as environment maps. In this paper, we present an alternative physically based approach that captures these effects while maintaining real time performance and the ease-of-use of the OpenGL fog model. Our method is based on an explicit analytic integration of the single scattering light transport equations for an isotropic point light source in a homogeneous participating medium. We can implement the model in modern programmable graphics hardware using a few small numerical lookup tables stored as texture maps. Our model can also be easily adapted to generate the appearances of materials with arbitrary BRDFs, environment map lighting, and precomputed radiance transfer methods, in the presence of participating media. Hence, our techniques can be widely used in real-time rendering.",
"Abstract In virtually all rendering systems, linear light sources are modeled with a series of point light sources that require considerable computing resources to produce realistic looking results. A general solution for shading surfaces illuminated by a linear light source is proposed. A formulation allowing for faster computation of the diffuse component of light reflection is derived. By assuming Phong's specular component, simple, inexpensive and convincing results are produced with the use of a Chebyshev approximation. A shadowing algorithm is also presented. As shadowing from linear light sources is expensive, two acceleration schemes, extended from ray tracing, are evaluated.",
"Although B-spline curves and surfaces have enjoyed a long established place in the graphics community as constructive modeling tools, the use of B-spline approximation techniques has received relatively little attention in rendering. In this work we explore the use of 4D and 5D tensor product B-spline functions to represent surface radiance, and establish that, when appropriately applied, they can be used effectively for static scenes with diffuse to moderately specular elements. Once computed, the surface radiance representation is view independent, can be evaluated quickly, and is equally suited for incorporation into ray tracing or scan-line rendering algorithms. Furthermore, we use B-spline approximation techniques to solve the problem of global illumination for general parametric surfaces with a wide range of reflectance and transmission properties. We conclude that addressing functional approximation aspects offers a fertile research ground relative to the already impressive gains that splines have made in other fields.",
"The irradiance at a point on a surface due to a polyhedral source of uniform brightness is given by a well-known analytic formula. In this paper we derive the corresponding analytic expression for the irradiance Jacobian , the derivative of the vector representation of irradiance. Although the result is elementary for unoccluded sources, within penumbrae the irradiance Jacobian must incorporate more information about blockers than either the irradiance or vector irradiance. The expression presented here holds for any number of polyhedral blockers and requires only a minor extension of standard polygon clipping to evaluate. To illustrate its use, three related applications are briefing described: direct computation of isolux contours, finding local irradiance extrema, and iso-meshing. Isolux contours are curves of constant irradiance across a surface that can be followed using a predictor-corrector method based on the irradiance Jacobian. Similarly, local extrema can be found using a descent method. Finally, iso-meshing is a new approach to surface mesh generation that incorporates families of isolux contours."
]
} |
1705.04927 | 2614901197 | In this paper, we present a new mathematical foundation for image-based lighting. Using a simple manipulation of the local coordinate system, we derive a closed-form solution to the light integral equation under distant environment illumination. We derive our solution for different BRDF's such as lambertian and Phong-like. The method is free of noise, and provides the possibility of using the full spectrum of frequencies captured by images taken from the environment. This allows for the color of the rendered object to be toned according to the color of the light in the environment. Experimental results also show that one can gain an order of magnitude or higher in rendering time compared to Monte Carlo quadrature methods and spherical harmonics. | Closed-form solutions were also proposed for special cases of non-constant lighting such as the work done by @cite_57 , which provides a solution for linearly-varying luminaires. The most common application for non-constant lighting is in environment maps. @cite_144 used spherical harmonics to solve the light integral for diffuse materials lit by environment maps, and achieved real-time rendering. | {
"cite_N": [
"@cite_57",
"@cite_144"
],
"mid": [
"1535548534",
"2105649179"
],
"abstract": [
"We present a closed-form expression for the irradiance at a point on a surface due to an arbitrary polygonal Lambertian luminaire with linearly-varying radiant exitance. The solution consists of elementary functions and a single well-behaved special function that can be either approximated directly or computed exactly in terms of classical special functions such as Clausen’s integral or the closely related dilogarithm. We first provide a general boundary integral that applies to all planar luminaires and then derive the closed-form expression that applies to arbitrary polygons, which is the result most relevant for global illumination. Our approach is to express the problem as an integral of a simple class of rational functions over regions of the sphere, and to convert the surface integral to a boundary integral using a generalization of irradiance tensors. The result extends the class of available closed-form expressions for computing direct radiative transfer from finite areas to differential areas. We provide an outline of the derivation, a detailed proof of the resulting formula, and complete pseudo-code of the resulting algorithm. Finally, we demonstrate the validity of our algorithm by comparison with Monte Carlo. While there are direct applications of this work, it is primarily of theoretical interest as it introduces much of the machinery needed to derive closed-form solutions for the general case of luminaires with radiance distributions that vary polynomially in both position and direction.",
"We consider the rendering of diffuse objects under distant illumination, as specified by an environment map. Using an analytic expression for the irradiance in terms of spherical harmonic coefficients of the lighting, we show that one needs to compute and use only 9 coefficients, corresponding to the lowest-frequency modes of the illumination, in order to achieve average errors of only 1 . In other words, the irradiance is insensitive to high frequencies in the lighting, and is well approximated using only 9 parameters. In fact, we show that the irradiance can be procedurally represented simply as a quadratic polynomial in the cartesian components of the surface normal, and give explicit formulae. These observations lead to a simple and efficient procedural rendering algorithm amenable to hardware implementation, a prefiltering method up to three orders of magnitude faster than previous techniques, and new representations for lighting design and image-based rendering."
]
} |
1705.04969 | 2950723285 | Embedding network data into a low-dimensional vector space has shown promising performance for many real-world applications, such as node classification and entity retrieval. However, most existing methods focused only on leveraging network structure. For social networks, besides the network structure, there also exists rich information about social actors, such as user profiles of friendship networks and textual content of citation networks. These rich attribute information of social actors reveal the homophily effect, exerting huge impacts on the formation of social networks. In this paper, we explore the rich evidence source of attributes in social networks to improve network embedding. We propose a generic Social Network Embedding framework (SNE), which learns representations for social actors (i.e., nodes) by preserving both the structural proximity and attribute proximity. While the structural proximity captures the global network structure, the attribute proximity accounts for the homophily effect. To justify our proposal, we conduct extensive experiments on four real-world social networks. Compared to the state-of-the-art network embedding approaches, SNE can learn more informative representations, achieving substantial gains on the tasks of link prediction and node classification. Specifically, SNE significantly outperforms node2vec with an 8.2 relative improvement on the link prediction task, and a 12.7 gain on the node classification task. | Some earlier works such as Local Linear Embedding (LLE) @cite_2 , IsoMAP @cite_12 and Laplacian Eigenmap @cite_25 first transform data into an affinity graph based on the feature vectors of nodes ( e.g., k-nearest neighbors of nodes) and then embed the graph by solving the leading eigenvectors of the affinity matrix. | {
"cite_N": [
"@cite_25",
"@cite_12",
"@cite_2"
],
"mid": [
"2156718197",
"",
"2053186076"
],
"abstract": [
"Drawing on the correspondence between the graph Laplacian, the Laplace-Beltrami operator on a manifold, and the connections to the heat equation, we propose a geometrically motivated algorithm for constructing a representation for data sampled from a low dimensional manifold embedded in a higher dimensional space. The algorithm provides a computationally efficient approach to nonlinear dimensionality reduction that has locality preserving properties and a natural connection to clustering. Several applications are considered.",
"",
"Many areas of science depend on exploratory data analysis and visualization. The need to analyze large amounts of multivariate data raises the fundamental problem of dimensionality reduction: how to discover compact representations of high-dimensional data. Here, we introduce locally linear embedding (LLE), an unsupervised learning algorithm that computes low-dimensional, neighborhood-preserving embeddings of high-dimensional inputs. Unlike clustering methods for local dimensionality reduction, LLE maps its inputs into a single global coordinate system of lower dimensionality, and its optimizations do not involve local minima. By exploiting the local symmetries of linear reconstructions, LLE is able to learn the global structure of nonlinear manifolds, such as those generated by images of faces or documents of text. How do we judge similarity? Our mental representations of the world are formed by processing large numbers of sensory in"
]
} |
1705.04969 | 2950723285 | Embedding network data into a low-dimensional vector space has shown promising performance for many real-world applications, such as node classification and entity retrieval. However, most existing methods focused only on leveraging network structure. For social networks, besides the network structure, there also exists rich information about social actors, such as user profiles of friendship networks and textual content of citation networks. These rich attribute information of social actors reveal the homophily effect, exerting huge impacts on the formation of social networks. In this paper, we explore the rich evidence source of attributes in social networks to improve network embedding. We propose a generic Social Network Embedding framework (SNE), which learns representations for social actors (i.e., nodes) by preserving both the structural proximity and attribute proximity. While the structural proximity captures the global network structure, the attribute proximity accounts for the homophily effect. To justify our proposal, we conduct extensive experiments on four real-world social networks. Compared to the state-of-the-art network embedding approaches, SNE can learn more informative representations, achieving substantial gains on the tasks of link prediction and node classification. Specifically, SNE significantly outperforms node2vec with an 8.2 relative improvement on the link prediction task, and a 12.7 gain on the node classification task. | Recent works focus more on embedding an existing network into a low-dimensional vector space to facilitate further analysis and achieve better performance than those earlier works. In @cite_1 the authors deployed truncated random walks on networks to generate node sequences. The generated node sequences are treated as sentences in language models and fed to the Skip-gram model to learn the embeddings. In @cite_32 the authors modified the way of generating node sequences by balancing breadth-first sampling and depth-first sampling, and achieved performance improvements. Instead of performing simulated walks'' on the networks, @cite_47 proposed clear objective functions to preserve the and of nodes while @cite_21 introduced deep models with multiple layers of non-linear functions to capture the highly non-linear network structure. However, all these methods only leverage network structure. In social networks, there exists large amount of attribute information. Purely structure-based methods fail to capture such valuable information, thus may result in less informative embeddings. In addition, these methods get affected easily when the link sparsity problem occurs. | {
"cite_N": [
"@cite_47",
"@cite_21",
"@cite_1",
"@cite_32"
],
"mid": [
"1888005072",
"2393319904",
"2154851992",
"2962756421"
],
"abstract": [
"This paper studies the problem of embedding very large information networks into low-dimensional vector spaces, which is useful in many tasks such as visualization, node classification, and link prediction. Most existing graph embedding methods do not scale for real world information networks which usually contain millions of nodes. In this paper, we propose a novel network embedding method called the LINE,'' which is suitable for arbitrary types of information networks: undirected, directed, and or weighted. The method optimizes a carefully designed objective function that preserves both the local and global network structures. An edge-sampling algorithm is proposed that addresses the limitation of the classical stochastic gradient descent and improves both the effectiveness and the efficiency of the inference. Empirical experiments prove the effectiveness of the LINE on a variety of real-world information networks, including language networks, social networks, and citation networks. The algorithm is very efficient, which is able to learn the embedding of a network with millions of vertices and billions of edges in a few hours on a typical single machine. The source code of the LINE is available online https: github.com tangjianpku LINE .",
"Network embedding is an important method to learn low-dimensional representations of vertexes in networks, aiming to capture and preserve the network structure. Almost all the existing network embedding methods adopt shallow models. However, since the underlying network structure is complex, shallow models cannot capture the highly non-linear network structure, resulting in sub-optimal network representations. Therefore, how to find a method that is able to effectively capture the highly non-linear network structure and preserve the global and local structure is an open yet important problem. To solve this problem, in this paper we propose a Structural Deep Network Embedding method, namely SDNE. More specifically, we first propose a semi-supervised deep model, which has multiple layers of non-linear functions, thereby being able to capture the highly non-linear network structure. Then we propose to exploit the first-order and second-order proximity jointly to preserve the network structure. The second-order proximity is used by the unsupervised component to capture the global network structure. While the first-order proximity is used as the supervised information in the supervised component to preserve the local network structure. By jointly optimizing them in the semi-supervised deep model, our method can preserve both the local and global network structure and is robust to sparse networks. Empirically, we conduct the experiments on five real-world networks, including a language network, a citation network and three social networks. The results show that compared to the baselines, our method can reconstruct the original network significantly better and achieves substantial gains in three applications, i.e. multi-label classification, link prediction and visualization.",
"We present DeepWalk, a novel approach for learning latent representations of vertices in a network. These latent representations encode social relations in a continuous vector space, which is easily exploited by statistical models. DeepWalk generalizes recent advancements in language modeling and unsupervised feature learning (or deep learning) from sequences of words to graphs. DeepWalk uses local information obtained from truncated random walks to learn latent representations by treating walks as the equivalent of sentences. We demonstrate DeepWalk's latent representations on several multi-label network classification tasks for social networks such as BlogCatalog, Flickr, and YouTube. Our results show that DeepWalk outperforms challenging baselines which are allowed a global view of the network, especially in the presence of missing information. DeepWalk's representations can provide F1 scores up to 10 higher than competing methods when labeled data is sparse. In some experiments, DeepWalk's representations are able to outperform all baseline methods while using 60 less training data. DeepWalk is also scalable. It is an online learning algorithm which builds useful incremental results, and is trivially parallelizable. These qualities make it suitable for a broad class of real world applications such as network classification, and anomaly detection.",
"Prediction tasks over nodes and edges in networks require careful effort in engineering features used by learning algorithms. Recent research in the broader field of representation learning has led to significant progress in automating prediction by learning the features themselves. However, present feature learning approaches are not expressive enough to capture the diversity of connectivity patterns observed in networks. Here we propose node2vec, an algorithmic framework for learning continuous feature representations for nodes in networks. In node2vec, we learn a mapping of nodes to a low-dimensional space of features that maximizes the likelihood of preserving network neighborhoods of nodes. We define a flexible notion of a node's network neighborhood and design a biased random walk procedure, which efficiently explores diverse neighborhoods. Our algorithm generalizes prior work which is based on rigid notions of network neighborhoods, and we argue that the added flexibility in exploring neighborhoods is the key to learning richer representations. We demonstrate the efficacy of node2vec over existing state-of-the-art techniques on multi-label classification and link prediction in several real-world networks from diverse domains. Taken together, our work represents a new way for efficiently learning state-of-the-art task-independent representations in complex networks."
]
} |
1705.04969 | 2950723285 | Embedding network data into a low-dimensional vector space has shown promising performance for many real-world applications, such as node classification and entity retrieval. However, most existing methods focused only on leveraging network structure. For social networks, besides the network structure, there also exists rich information about social actors, such as user profiles of friendship networks and textual content of citation networks. These rich attribute information of social actors reveal the homophily effect, exerting huge impacts on the formation of social networks. In this paper, we explore the rich evidence source of attributes in social networks to improve network embedding. We propose a generic Social Network Embedding framework (SNE), which learns representations for social actors (i.e., nodes) by preserving both the structural proximity and attribute proximity. While the structural proximity captures the global network structure, the attribute proximity accounts for the homophily effect. To justify our proposal, we conduct extensive experiments on four real-world social networks. Compared to the state-of-the-art network embedding approaches, SNE can learn more informative representations, achieving substantial gains on the tasks of link prediction and node classification. Specifically, SNE significantly outperforms node2vec with an 8.2 relative improvement on the link prediction task, and a 12.7 gain on the node classification task. | Some recent efforts have explored the possibility of integrating contents to learn better representations @cite_23 . For example, @cite_3 proposed text-associated @cite_1 to incorporate text features into the matrix factorization framework. However, only text attributes can be handled. Being with the same problem, @cite_27 proposed to separately learn embeddings from the structure-based @cite_1 and label-fused model @cite_10 , the embeddings learned were linearly combined together in an iterative way. Under such a scheme, the knowledge interaction between the two separate models only goes through a series of weighted sum operations and lacks further convergence constrains. On the contrary, our method models the structure proximity and attribute proximity in an end-to-end neural network that does not have such limitations. Also, by incorporating structure and attribute modeling by an early fusion, the two parts only need to complement each other, resulting in sufficient knowledge interactions @cite_18 . | {
"cite_N": [
"@cite_18",
"@cite_1",
"@cite_3",
"@cite_27",
"@cite_23",
"@cite_10"
],
"mid": [
"",
"2154851992",
"",
"2574817444",
"2583803680",
"2131744502"
],
"abstract": [
"",
"We present DeepWalk, a novel approach for learning latent representations of vertices in a network. These latent representations encode social relations in a continuous vector space, which is easily exploited by statistical models. DeepWalk generalizes recent advancements in language modeling and unsupervised feature learning (or deep learning) from sequences of words to graphs. DeepWalk uses local information obtained from truncated random walks to learn latent representations by treating walks as the equivalent of sentences. We demonstrate DeepWalk's latent representations on several multi-label network classification tasks for social networks such as BlogCatalog, Flickr, and YouTube. Our results show that DeepWalk outperforms challenging baselines which are allowed a global view of the network, especially in the presence of missing information. DeepWalk's representations can provide F1 scores up to 10 higher than competing methods when labeled data is sparse. In some experiments, DeepWalk's representations are able to outperform all baseline methods while using 60 less training data. DeepWalk is also scalable. It is an online learning algorithm which builds useful incremental results, and is trivially parallelizable. These qualities make it suitable for a broad class of real world applications such as network classification, and anomaly detection.",
"",
"Information network mining often requires examination of linkage relationships between nodes for analysis. Recently, network representation has emerged to represent each node in a vector format, embedding network structure, so off-the-shelf machine learning methods can be directly applied for analysis. To date, existing methods only focus on one aspect of node information and cannot leverage node labels. In this paper, we propose TriDNR, a tri-party deep network representation model, using information from three parties: node structure, node content, and node labels (if available) to jointly learn optimal node representation. TriDNR is based on our new coupled deep natural language module, whose learning is enforced at three levels: (1) at the network structure level, TriDNR exploits inter-node relationship by maximizing the probability of observing surrounding nodes given a node in random walks; (2) at the node content level, TriDNR captures node-word correlation by maximizing the co-occurrence of word sequence given a node; and (3) at the node label level, TriDNR models label-word correspondence by maximizing the probability of word sequence given a class label. The tri-party information is jointly fed into the neural network model to mutually enhance each other to learn optimal representation, and results in up to 79 classification accuracy gain, compared to state-of-the-art methods.",
"Attributed network embedding aims to seek low-dimensional vector representations for nodes in a network, such that original network topological structure and node attribute proximity can be preserved in the vectors. These learned representations have been demonstrated to be helpful in many learning tasks such as network clustering and link prediction. While existing algorithms follow an unsupervised manner, nodes in many real-world attributed networks are often associated with abundant label information, which is potentially valuable in seeking more effective joint vector representations. In this paper, we investigate how labels can be modeled and incorporated to improve attributed network embedding. This is a challenging task since label information could be noisy and incomplete. In addition, labels are completely distinct with the geometrical structure and node attributes. The bewildering combination of heterogeneous information makes the joint vector representation learning more difficult. To address these issues, we propose a novel Label informed Attributed Network Embedding (LANE) framework. It can smoothly incorporate label information into the attributed network embedding while preserving their correlations. Experiments on real-world datasets demonstrate that the proposed framework achieves significantly better performance compared with the state-of-the-art embedding algorithms.",
"Many machine learning algorithms require the input to be represented as a fixed-length feature vector. When it comes to texts, one of the most common fixed-length features is bag-of-words. Despite their popularity, bag-of-words features have two major weaknesses: they lose the ordering of the words and they also ignore semantics of the words. For example, \"powerful,\" \"strong\" and \"Paris\" are equally distant. In this paper, we propose Paragraph Vector, an unsupervised algorithm that learns fixed-length feature representations from variable-length pieces of texts, such as sentences, paragraphs, and documents. Our algorithm represents each document by a dense vector which is trained to predict words in the document. Its construction gives our algorithm the potential to overcome the weaknesses of bag-of-words models. Empirical results show that Paragraph Vectors outperforms bag-of-words models as well as other techniques for text representations. Finally, we achieve new state-of-the-art results on several text classification and sentiment analysis tasks."
]
} |
1705.05087 | 2953004454 | With ever-increasing productivity targets in mining operations, there is a growing interest in mining automation. In future mines, remote-controlled and autonomous haulers will operate underground guided by LiDAR sensors. We envision reusing LiDAR measurements to maintain accurate mine maps that would contribute to both safety and productivity. Extrapolating from a pilot project on reliable wireless communication in Boliden's Kankberg mine, we propose establishing a system-of-systems (SoS) with LIDAR-equipped haulers and existing mapping solutions as constituent systems. SoS requirements engineering inevitably adds a political layer, as independent actors are stakeholders both on the system and SoS levels. We present four SoS scenarios representing different business models, discussing how development and operations could be distributed among Boliden and external stakeholders, e.g., the vehicle suppliers, the hauling company, and the developers of the mapping software. Based on eight key variation points, we compare the four scenarios from both technical and business perspectives. Finally, we validate our findings in a seminar with participants from the relevant stakeholders. We conclude that to determine which scenario is the most promising for Boliden, trade-offs regarding control, costs, risks, and innovation must be carefully evaluated. | Boliden has an explicit ambition to run projects in-house, i.e., internal know-how is considered fundamental. The current automation trend promises increased productivity through the introduction of autonomous machines, wireless data transfer, and positioning of people and equipment –- all these solutions rely on software-intensive systems. However, whether Boliden will be able to keep all required technical software expertise in-house is uncertain. The scaling role of software in traditional industries has attracted considerable research efforts lately @cite_12 , and turning into a software-intensive company is an acknowledged challenge. The IT department at Boliden roughly employs 100 people including support functions, but few of them are software developers. | {
"cite_N": [
"@cite_12"
],
"mid": [
"2782386751"
],
"abstract": [
"This book is open access under a CC BY 4.0 license. This book is intended primarily for practitioners who are facing the softwareisation of their business. It presents the Scaling Management Framework, a model based on collected experiences from companies that have already made the journey to give software a central role within the organization. The model is unique because it suggests a holistic method to analyze and plan your journey. It simply means that you cant just focus solely on your products or services. You also have to look closely at your processes and your organization, the way you make decisions and get things done. Inevitably, these will have to change. Software has changed the rules of the game. The world talks about the digitalization in industry and society how the focus has shifted from producing tangible things towards software and services. This trend started many years ago, but is now affecting every company, whether its a software company or not. There are many companies that have already made a digitalization journey and many are about to embark on this journey like you. How do you transform your organization when software is becoming a critical part of your business? This book comes with a map, a compass, and suggested journeys along with selected travel stories comprising best practices and lessons learned from past digitalization journeys. Use the map to find your way in the digitalization landscape, and use the compass to find the direction of your journey."
]
} |
1705.05087 | 2953004454 | With ever-increasing productivity targets in mining operations, there is a growing interest in mining automation. In future mines, remote-controlled and autonomous haulers will operate underground guided by LiDAR sensors. We envision reusing LiDAR measurements to maintain accurate mine maps that would contribute to both safety and productivity. Extrapolating from a pilot project on reliable wireless communication in Boliden's Kankberg mine, we propose establishing a system-of-systems (SoS) with LIDAR-equipped haulers and existing mapping solutions as constituent systems. SoS requirements engineering inevitably adds a political layer, as independent actors are stakeholders both on the system and SoS levels. We present four SoS scenarios representing different business models, discussing how development and operations could be distributed among Boliden and external stakeholders, e.g., the vehicle suppliers, the hauling company, and the developers of the mapping software. Based on eight key variation points, we compare the four scenarios from both technical and business perspectives. Finally, we validate our findings in a seminar with participants from the relevant stakeholders. We conclude that to determine which scenario is the most promising for Boliden, trade-offs regarding control, costs, risks, and innovation must be carefully evaluated. | As the mining operations proceed, shafts and drifts are inevitably affected in position (inclination, rotation, lateral movements, and curves) and form (compression and deformation). There are three main causes of cave-ins in underground mines. First, hasty mining operations might fail to secure walls and ceilings of shafts and drifts. Second, excessive excavation might lead to cracks in floors and walls, thus weakening the entire structure. Examples include insufficient vertical spacing between crosscuts and too rectangular crosscuts causing stress concentration in corners. Third, gradual sinking of land can cause cave-ins @cite_20 , i.e., subsidence (the downward motion of a surface). As illustrated in Fig. , mining operations induce subsidence of the Earth's surface. While mining-induced subsidence is rather predictable in magnitude and extent, monitoring the progress is fundamental to mining safety. In Kankberg, however, the mountain stresses caused by drilling and blasting activities dominate any subsidence. | {
"cite_N": [
"@cite_20"
],
"mid": [
"2066713014"
],
"abstract": [
"This article discusses the software architecture of an autonomous robotic system designed to explore and map abandoned mines. A new set of software tools is presented, enabling robots to acquire maps of unprecedented size and accuracy. On 30 May 2003, the robot \"Groundhog\" successfully explored and mapped a main corridor of the abandoned Mathies mine near Courtney, Pennsylvania. This article also discusses some of the challenges that arise in the subterranean environments and some the difficulties of building truly autonomous robots."
]
} |
1705.05087 | 2953004454 | With ever-increasing productivity targets in mining operations, there is a growing interest in mining automation. In future mines, remote-controlled and autonomous haulers will operate underground guided by LiDAR sensors. We envision reusing LiDAR measurements to maintain accurate mine maps that would contribute to both safety and productivity. Extrapolating from a pilot project on reliable wireless communication in Boliden's Kankberg mine, we propose establishing a system-of-systems (SoS) with LIDAR-equipped haulers and existing mapping solutions as constituent systems. SoS requirements engineering inevitably adds a political layer, as independent actors are stakeholders both on the system and SoS levels. We present four SoS scenarios representing different business models, discussing how development and operations could be distributed among Boliden and external stakeholders, e.g., the vehicle suppliers, the hauling company, and the developers of the mapping software. Based on eight key variation points, we compare the four scenarios from both technical and business perspectives. Finally, we validate our findings in a seminar with participants from the relevant stakeholders. We conclude that to determine which scenario is the most promising for Boliden, trade-offs regarding control, costs, risks, and innovation must be carefully evaluated. | Huber and Vandapel and their research group also did work on underground mine mapping using robots @cite_18 . In contrast to the work by Thrun , Huber and Vandapel tried their approach in active mines. They mounted a high-resolution 3D scanner on a mobile robot, providing 8000 x 1400 pixel scans with millimeter-level accuracy. As for Groundhog, they collected scans only at certain vantage points; Each three to five meters the robot stopped for 90 seconds to obtain a complete scan. A considerable contribution of their research is related to multi-view surface matching, i.e., merging multiple 3D views into a single map. Their approach is called iterative merging, which was successfully used to create high-quality maps of an underground mine. However, their approach does not scale to large numbers of scans; Back in 2006, their approach could only generate sub-maps containing about 50 scans. | {
"cite_N": [
"@cite_18"
],
"mid": [
"2888726385"
],
"abstract": [
"For several years, our research group has been developing methods for automated modeling of 3D environments. In September, 2002, we were given the opportunity to demonstrate our mapping capability in an underground coal mine. The opportunity arose as a result of the Quecreek mine accident, in which an inaccurate map caused miners to breach an abandoned, water-lled mine, trapping them for several days. Our eld test illustrates the feasibility and potential of high resolution three-dimensional (3D) mapping of an underground coal mine using a cartmounted 3D laser scanner. This paper presents our experimental setup, the automatic 3D modeling method used, and the results of the eld test. In addition, we address issues related to laser sensing in a coal mine environment."
]
} |
1705.05138 | 2736021506 | Scalar features in time-dependent fluid flow are traditionally visualized using 3D representation, and their topology changes over time are often conveyed with abstract graphs. Using such techniques, however, the structural details of feature separation and the temporal evolution of features undergoing topological changes are difficult to analyze. In this paper, we propose a novel approach for the spatio-temporal visualization of feature separation that segments feature volumes into regions with respect to their contribution to distinct features after separation. To this end, we employ particle-based feature tracking to find volumetric correspondences between features at two different instants of time. We visualize this segmentation by constructing mesh boundaries around each volume segment of a feature at the initial time that correspond to the separated features at the later time. To convey temporal evolution of the partitioning within the investigated time interval, we complement our approach with spatio-temporal separation surfaces. For the application of our approach to multiphase flow, we additionally present a feature-based corrector method to ensure phase-consistent particle trajectories. The utility of our technique is demonstrated by application to two-phase (liquid-gas) and multi-component (liquid-liquid) flows where the scalar field represents the fraction of one of the phases. | For the visualization in multiphase flow, we use a corrector method to ensure that the advected particles remain in the given phase during tracing. Related to this problem are approaches for surface tracking. Stam @cite_30 proposed a method for the computation of interface velocities to properly translate surfaces. Bojsen- @cite_31 developed a method for tracking surfaces undergoing topology changes, without prior information on the underlying physics. In the work by @cite_24 , Lagrangian transport is used for correct displacement interpolation. @cite_6 optimize the transportation in terms of Wasserstein distances, which allows for efficient shape transformation. In our work, we analyze the evolution of volumes, and computing Wasserstein distances in this case would be computationally prohibitive and also would not guarantee physically correct correspondences. | {
"cite_N": [
"@cite_30",
"@cite_31",
"@cite_6",
"@cite_24"
],
"mid": [
"2109760112",
"1967167311",
"",
"2029241756"
],
"abstract": [
"In this article we derive an equation for the velocity of an arbitrary time-evolving implicit surface. Strictly speaking, only the normal component of the velocity is unambiguously defined. This is because an implicit surface does not have a unique parametrization. However, by enforcing a constraint on the evolution of the normal field we obtain a unique tangential component. We apply our formulas to surface tracking and to the problem of computing velocity vectors of a motion blurred blobby surface. Other possible applications are mentioned at the end of the article.",
"We present a method for recovering a temporally coherent, deforming triangle mesh with arbitrarily changing topology from an incoherent sequence of static closed surfaces. We solve this problem using the surface geometry alone, without any prior information like surface templates or velocity fields. Our system combines a proven strategy for triangle mesh improvement, a robust multi-resolution non-rigid registration routine, and a reliable technique for changing surface mesh topology. We also introduce a novel topological constraint enforcement algorithm to ensure that the output and input always have similar topology. We apply our technique to a series of diverse input data from video reconstructions, physics simulations, and artistic morphs. The structured output of our algorithm allows us to efficiently track information like colors and displacement maps, recover velocity information, and solve PDEs on the mesh as a post process.",
"",
"Interpolation between pairs of values, typically vectors, is a fundamental operation in many computer graphics applications. In some cases simple linear interpolation yields meaningful results without requiring domain knowledge. However, interpolation between pairs of distributions or pairs of functions often demands more care because features may exhibit translational motion between exemplars. This property is not captured by linear interpolation. This paper develops the use of displacement interpolation for this class of problem, which provides a generic method for interpolating between distributions or functions based on advection instead of blending. The functions can be non-uniformly sampled, high-dimensional, and defined on non-Euclidean manifolds, e.g., spheres and tori. Our method decomposes distributions or functions into sums of radial basis functions (RBFs). We solve a mass transport problem to pair the RBFs and apply partial transport to obtain the interpolated function. We describe practical methods for computing the RBF decomposition and solving the transport problem. We demonstrate the interpolation approach on synthetic examples, BRDFs, color distributions, environment maps, stipple patterns, and value functions."
]
} |
1705.05137 | 2950112792 | CSPe is a specification language for runtime monitors that can directly express concurrency in a bottom-up manner that composes the system from simpler, interacting components. It includes constructs to explicitly flag failures to the monitor, which unlike deadlocks and livelocks in conventional process algebras, propagate globally and aborts the whole system's execution. Although CSPe has a trace semantics along with an implementation demonstrating acceptable performance, it lacks an operational semantics. An operational semantics is not only more accessible than trace semantics but also indispensable for ensuring the correctness of the implementation. Furthermore, a process algebra like CSPe admits multiple denotational semantics appropriate for different purposes, and an operational semantics is the basis for justifying such semantics' integrity and relevance. In this paper, we develop an SOS-style operational semantics for CSPe, which properly accounts for explicit failures and will serve as a basis for further study of its properties, its optimization, and its use in runtime verification. | The main issue with semantics is the propagation of @math , which entails the negative constraint that normal computation rules apply only if @math -propagation rules do not. Negative premises of the form @math come quite naturally as a means for codifying such constraints, but negative premises are generally quite problematic. A transition relation satisfying negative rules may be not-existent, or non-unique, with no obvious guiding principle (such as minimality) in choosing the right'' one. Some formats do guarantee well-definedness, such as GSOS with the witnessing constraint @cite_2 and @cite_7 . But even then, negative rules tend to betray desirable properties such as compositionality of some forms of bisimulation @cite_4 . | {
"cite_N": [
"@cite_4",
"@cite_7",
"@cite_2"
],
"mid": [
"2041561080",
"",
"1966112122"
],
"abstract": [
"In this study, we present rule formats for four main notions of bisimulation with silent moves. Weak bisimulation is a congruence for any process algebra defined by WB cool rules; we have similar results for rooted weak bisimulation (Milner''s observational equivalence''''), branching bisimulation, and rooted branching bisimulation. The theorems stating that, say, observational equivalence is an appropriate notion of equality for CCS are corollaries of the results of this paper. We also give sufficient conditions under which equational axiom systems can be generated from operational rules. Indeed, many equational axiom systems appearing in the literature are instances of this general theory.",
"",
"In the concurrent language CCS, two programs are considered the same if they are bisimilar . Several years and many researchers have demonstrated that the theory of bisimulation is mathematically appealing and useful in practice. However, bisimulation makes too many distinctions between programs. We consider the problem of adding operations to CCS to make bisimulation fully abstract. We define the class of GSOS operations, generalizing the style and technical advantages of CCS operations. We characterize GSOS congruence in as a bisimulation-like relation called ready-simulation . Bisimulation is strictly finer than ready simulation, and hence not a congruence for any GSOS language."
]
} |
1705.05137 | 2950112792 | CSPe is a specification language for runtime monitors that can directly express concurrency in a bottom-up manner that composes the system from simpler, interacting components. It includes constructs to explicitly flag failures to the monitor, which unlike deadlocks and livelocks in conventional process algebras, propagate globally and aborts the whole system's execution. Although CSPe has a trace semantics along with an implementation demonstrating acceptable performance, it lacks an operational semantics. An operational semantics is not only more accessible than trace semantics but also indispensable for ensuring the correctness of the implementation. Furthermore, a process algebra like CSPe admits multiple denotational semantics appropriate for different purposes, and an operational semantics is the basis for justifying such semantics' integrity and relevance. In this paper, we develop an SOS-style operational semantics for CSPe, which properly accounts for explicit failures and will serve as a basis for further study of its properties, its optimization, and its use in runtime verification. | Our approach exploits the fact that we only have a very specific negative constraint -- the absence of doomed subprocesses -- and encodes it with a restriction on the range of metavariables in transition rules. With trick, we manage to avoid negative premises altogether, essentially turning the system into a positive one. This approach is very commonly employed, e.g. in reduction rules for the call-by-value @math calculus @cite_8 , where the argument in a function application should be evaluated only if the function expression cannot be evaluated any further. | {
"cite_N": [
"@cite_8"
],
"mid": [
"2084788336"
],
"abstract": [
"Part 1 Introduction: model programming languages lambda notation equations, reduction and semantics types and type systems notation and mathematical conventions set-theoretic background syntax and semantics induction. Part 2 The language PCF: syntax of PCF PCF programmes and their semantics PCF reduction and symbolic interpreters PCF programming examples, expressive power and limitations variations and extensions of PCF. Part 3 Universal algebra and algebraic data types: preview of algebraic specification algebras, signatures and terms equations, soundness and completeness homomorphisms and initiality algebraic data types rewrite systems. Part 4 Simply-typed lambda calculus: types terms proof systems Henkin models, soundness and completeness. Part 5 Models of typed lambda calculus: domain-theoretic models and fixed points fixed-point induction computational adequacy and full abstraction recursion-theoretic models partial equivalence relations and recursion. Part 6 Imperative programmes: while programmes operational semantics denotational semantics before-after assertions about while programmes semantics of additional programme constructs. Part 7 Categories and recursive types: Cartesian closed categories Kripke lambda models and functor categories domain models of recursive types. Part 8 Logical relations: introduction to logical relations logical relations over applicative structures proof-theoretic results partial surjections and specific models representation independence generalizations of logical relations. Part 9 Polymorphism and modularity: predicative polymorphic calculus impredicative polymorphism data abstraction and existential types general products, sums and programme modules. Part 10 subtyping and related concepts: simply typed lambda calculus with subtyping records, semantic models of subtyping recursive types and a record model of objects polymorphism with subtype constraints. Part 11 Type inference: introduction to type inference type inference for lambda xxx with type variables type inference with polymorphic declarations."
]
} |
1705.05137 | 2950112792 | CSPe is a specification language for runtime monitors that can directly express concurrency in a bottom-up manner that composes the system from simpler, interacting components. It includes constructs to explicitly flag failures to the monitor, which unlike deadlocks and livelocks in conventional process algebras, propagate globally and aborts the whole system's execution. Although CSPe has a trace semantics along with an implementation demonstrating acceptable performance, it lacks an operational semantics. An operational semantics is not only more accessible than trace semantics but also indispensable for ensuring the correctness of the implementation. Furthermore, a process algebra like CSPe admits multiple denotational semantics appropriate for different purposes, and an operational semantics is the basis for justifying such semantics' integrity and relevance. In this paper, we develop an SOS-style operational semantics for CSPe, which properly accounts for explicit failures and will serve as a basis for further study of its properties, its optimization, and its use in runtime verification. | We identify @math -induced failures by transitions into @math , but an alternative approach would be to have @math emit a special event @math , just as termination is signalled by @math . Though we have not pursued this idea in detail, the central concern there will be to give @math higher priority than all other events. Prioritized transition also involves a negative constraint but is known to be quite well-behaved, being translatable to plain CSP @cite_6 . At the moment, it is not clear if @math propagation can be translated to the prioritized-transition primitive in @cite_6 . | {
"cite_N": [
"@cite_6"
],
"mid": [
"1793582516"
],
"abstract": [
"The author previously A.W. Roscoe, On the expressiveness of CSP, https: www.cs.ox.ac.uk files 1383 expressive.pdf, 2011; A.W. Roscoe, Understanding concurrent systems, Springer 2010 defined CSP-like operational semantics whose main restrictions were the automatic promotion of most ? actions, no cloning of running processes, and no negative premises in operational semantic rules. He showed that every operator with such an operational semantics can be translated into CSP and therefore has a semantics in every model of CSP. In this paper we demonstrate that a similar result holds for CSP extended by the priority operator described in Chapter 20 of A.W. Roscoe, Understanding concurrent systems, Springer 2010, with the restriction on negative premises removed."
]
} |
1705.05170 | 2615843232 | The software powering today's vehicles surpasses mechatronics as the dominating engineering challenge due to its fast evolving and innovative nature. In addition, the software and system architecture for upcoming vehicles with automated driving functionality is already processing 750MB s - corresponding to over 180 simultaneous 4K-video streams from popular video-on-demand services. Hence, self-driving cars will run so much software to resemble "small data centers on wheels" rather than just transportation vehicles. Continuous Integration, Deployment, and Experimentation have been successfully adopted for software-only products as enabling methodology for feedback-based software development. For example, a popular search engine conducts 250 experiments each day to improve the software based on its users' behavior. This work investigates design criteria for the software architecture and the corresponding software development and deployment process for complex cyber-physical systems, with the goal of enabling Continuous Experimentation as a way to achieve continuous software evolution. Our research involved reviewing related literature on the topic to extract relevant design requirements. The study is concluded by describing the software development and deployment process and software architecture adopted by our self-driving vehicle laboratory, both based on the extracted criteria. | One first relevant work is Baker and Dolan @cite_19 , that describes the software powering in the autonomous car Boss'', the vehicle that won the 2007 DARPA Urban Challenge. In this paper it is outlined how the software was built following the pattern, which works in a conceptually similar way as the more common pattern, providing both module inter-communication and decoupling. | {
"cite_N": [
"@cite_19"
],
"mid": [
"2120581524"
],
"abstract": [
"We describe an autonomous robotic software subsystem for managing mission execution and discrete traffic interaction in the 2007 DARPA Urban Challenge. Its role is reviewed in the context of the software system that controls ldquoBossrdquo, Tartan Racingpsilas winning entry in the competition. Design criteria are presented, followed by the application of software design principles to derive an architecture well suited to the rigors of developing complex robotic systems. Combined with a discussion of robust behavioral algorithms, the designpsilas effectiveness is highlighted in its ability to manage complex autonomous driving behaviors while remaining adaptable to the systempsilas evolving capabilities."
]
} |
1705.05170 | 2615843232 | The software powering today's vehicles surpasses mechatronics as the dominating engineering challenge due to its fast evolving and innovative nature. In addition, the software and system architecture for upcoming vehicles with automated driving functionality is already processing 750MB s - corresponding to over 180 simultaneous 4K-video streams from popular video-on-demand services. Hence, self-driving cars will run so much software to resemble "small data centers on wheels" rather than just transportation vehicles. Continuous Integration, Deployment, and Experimentation have been successfully adopted for software-only products as enabling methodology for feedback-based software development. For example, a popular search engine conducts 250 experiments each day to improve the software based on its users' behavior. This work investigates design criteria for the software architecture and the corresponding software development and deployment process for complex cyber-physical systems, with the goal of enabling Continuous Experimentation as a way to achieve continuous software evolution. Our research involved reviewing related literature on the topic to extract relevant design requirements. The study is concluded by describing the software development and deployment process and software architecture adopted by our self-driving vehicle laboratory, both based on the extracted criteria. | Another related work focusing on the architectural level was discussed in our previous work @cite_13 , where the results of a systematic literature review and a multiple case study are presented. The authors summarized their findings highlighting the following key aspects that characterized the resource-constrained system and software architectures for self-driving vehicles they analyzed: | {
"cite_N": [
"@cite_13"
],
"mid": [
"2397446945"
],
"abstract": [
"Context: Self-Driving cars are getting more and more attention with public demonstration from all important automotive OEMs but also from companies, which do not have a long history in the automotive industry. Fostered by large international competitions in the last decade, several automotive OEMs have already announced to bring this technology to the market around 2020. Objective: International competitions like the 2007 DARPA Urban Challenge did not focus on efficient usage of resources to realize the self-driving vehicular functionality. Since the automotive industry is very cost-sensitive, realizing reliable and robust self- driving functionality is challenging when expensive and sophisticated sensors mounted very visibly on the vehicle’s roof for example cannot be used. Therefore, the goal for this study is to investigate how architectural design decisions of recent self-driving vehicular technology consider resource-efficiency. Method: In a multiple case study, the architectural design decisions derived for resource- constrained self-driving miniature cars for the international competition CaroloCup are compared with architectural designs from recent real-scale self-driving cars. Results: Scaling down available resources for realizing self-driving vehicular technol- ogy puts additional constraints on the architectural design; especially reusability of software components in platform-independent algorithmic concepts are prevailing. Conclusion: Software frameworks like the robotic operating system (ROS) enable fast prototypical solutions; however, architectural support for resource-constrained devices is limited. Here, architectural design drivers as realized in AUTOSAR are more suitable."
]
} |
1705.05102 | 2614174454 | Image search and retrieval engines rely heavily on textual annotation in order to match word queries to a set of candidate images. A system that can automatically annotate images with meaningful text can be highly beneficial for such engines. Currently, the approaches to develop such systems try to establish relationships between keywords and visual features of images. In this paper, We make three main contributions to this area: (i) We transform this problem from the low-level keyword space to the high-level semantics space that we refer to as the " image theme ", (ii) Instead of treating each possible keyword independently, we use latent Dirichlet allocation to learn image themes from the associated texts in a training phase. Images are then annotated with image themes rather than keywords, using a modified continuous relevance model, which takes into account the spatial coherence and the visual continuity among images of common theme. (iii) To achieve more coherent annotations among images of common theme, we have integrated ConceptNet in learning the semantics of images, and hence augment image descriptions beyond annotations provided by humans. Images are thus further annotated by a few most significant words of the prominent image theme. Our extensive experiments show that a coherent theme-based image annotation using high-level semantics results in improved precision and recall as compared with equivalent classical keyword annotation systems. | The idea of annotating images with keywords has been vastly studied using different approaches in the literature, with most papers using the Corel 5K dataset as their benchmark @cite_76 @cite_8 @cite_102 . All different approaches essentially try to learn the relationships between words and image features. have provided a comprehensive review of all the popular techniques used for automatic image annotation @cite_54 . | {
"cite_N": [
"@cite_54",
"@cite_76",
"@cite_102",
"@cite_8"
],
"mid": [
"",
"1666447063",
"2127411609",
"2137918516"
],
"abstract": [
"",
"We describe a model of object recognition as machine translation. In this model, recognition is a process of annotating image regions with words. Firstly, images are segmented into regions, which are classified into region types using a variety of features. A mapping between region types and keywords supplied with the images, is then learned, using a method based around EM. This process is analogous with learning a lexicon from an aligned bitext. For the implementation we describe, these words are nouns taken from a large vocabulary. On a large test set, the method can predict numerous words with high accuracy. Simple methods identify words that cannot be predicted well. We show how to cluster words that individually are difficult to predict into clusters that can be predicted well -- for example, we cannot predict the distinction between train and locomotive using the current set of features, but we can predict the underlying concept. The method is trained on a substantial collection of images. Extensive experimental results illustrate the strengths and weaknesses of the approach.",
"We propose an approach to learning the semantics of images which allows us to automatically annotate an image with keywords and to retrieve images based on text queries. We do this using a formalism that models the generation of annotated images. We assume that every image is divided into regions, each described by a continuous-valued feature vector. Given a training set of images with annotations, we compute a joint probabilistic model of image features and words which allow us to predict the probability of generating a word given the image regions. This may be used to automatically annotate and retrieve images given a word as a query. Experiments show that our model significantly outperforms the best of the previously reported results on the tasks of automatic image annotation and retrieval.",
"Libraries have traditionally used manual image annotation for indexing and then later retrieving their image collections. However, manual image annotation is an expensive and labor intensive procedure and hence there has been great interest in coming up with automatic ways to retrieve images based on content. Here, we propose an automatic approach to annotating and retrieving images based on a training set of images. We assume that regions in an image can be described using a small vocabulary of blobs. Blobs are generated from image features using clustering. Given a training set of images with annotations, we show that probabilistic models allow us to predict the probability of generating a word given the blobs in an image. This may be used to automatically annotate and retrieve images given a word as a query. We show that relevance models allow us to derive these probabilities in a natural way. Experiments show that the annotation performance of this cross-media relevance model is almost six times as good (in terms of mean precision) than a model based on word-blob co-occurrence model and twice as good as a state of the art model derived from machine translation. Our approach shows the usefulness of using formal information retrieval models for the task of image annotation and retrieval."
]
} |
1705.05102 | 2614174454 | Image search and retrieval engines rely heavily on textual annotation in order to match word queries to a set of candidate images. A system that can automatically annotate images with meaningful text can be highly beneficial for such engines. Currently, the approaches to develop such systems try to establish relationships between keywords and visual features of images. In this paper, We make three main contributions to this area: (i) We transform this problem from the low-level keyword space to the high-level semantics space that we refer to as the " image theme ", (ii) Instead of treating each possible keyword independently, we use latent Dirichlet allocation to learn image themes from the associated texts in a training phase. Images are then annotated with image themes rather than keywords, using a modified continuous relevance model, which takes into account the spatial coherence and the visual continuity among images of common theme. (iii) To achieve more coherent annotations among images of common theme, we have integrated ConceptNet in learning the semantics of images, and hence augment image descriptions beyond annotations provided by humans. Images are thus further annotated by a few most significant words of the prominent image theme. Our extensive experiments show that a coherent theme-based image annotation using high-level semantics results in improved precision and recall as compared with equivalent classical keyword annotation systems. | Relevance models from machine translation were introduced to solve this problem by @cite_8 . To apply the relevance models, it is necessary to represent images in terms of visual features in a manner similar to the way documents are represented in terms of word-counts. Therefore, and many other researchers used the bag-of-words approach for image representation, which clusters image features to produce a finite number of visual-words. Blobworld by was popularly used for dividing images into meaningful patches of similar color and texture @cite_119 . introduced the relevance model in the continuous space named as the continuous-relevance-model (CRM) @cite_102 , and showed considerable improvement by removing the constraint of finite number of visual-words. introduced the multiple-bernoulli-relevance model and observed that dividing images into a fixed size grid works better than the complex system of Blobworld @cite_24 . | {
"cite_N": [
"@cite_24",
"@cite_119",
"@cite_102",
"@cite_8"
],
"mid": [
"2156336347",
"2135705692",
"2127411609",
"2137918516"
],
"abstract": [
"Retrieving images in response to textual queries requires some knowledge of the semantics of the picture. Here, we show how we can do both automatic image annotation and retrieval (using one word queries) from images and videos using a multiple Bernoulli relevance model. The model assumes that a training set of images or videos along with keyword annotations is provided. Multiple keywords are provided for an image and the specific correspondence between a keyword and an image is not provided. Each image is partitioned into a set of rectangular regions and a real-valued feature vector is computed over these regions. The relevance model is a joint probability distribution of the word annotations and the image feature vectors and is computed using the training set. The word probabilities are estimated using a multiple Bernoulli model and the image feature probabilities using a non-parametric kernel density estimate. The model is then used to annotate images in a test set. We show experiments on both images from a standard Corel data set and a set of video key frames from NIST's video tree. Comparative experiments show that the model performs better than a model based on estimating word probabilities using the popular multinomial distribution. The results also show that our model significantly outperforms previously reported results on the task of image and video annotation.",
"Retrieving images from large and varied collections using image content as a key is a challenging and important problem. We present a new image representation that provides a transformation from the raw pixel data to a small set of image regions that are coherent in color and texture. This \"Blobworld\" representation is created by clustering pixels in a joint color-texture-position feature space. The segmentation algorithm is fully automatic and has been run on a collection of 10,000 natural images. We describe a system that uses the Blobworld representation to retrieve images from this collection. An important aspect of the system is that the user is allowed to view the internal representation of the submitted image and the query results. Similar systems do not offer the user this view into the workings of the system; consequently, query results from these systems can be inexplicable, despite the availability of knobs for adjusting the similarity metrics. By finding image regions that roughly correspond to objects, we allow querying at the level of objects rather than global image properties. We present results indicating that querying for images using Blobworld produces higher precision than does querying using color and texture histograms of the entire image in cases where the image contains distinctive objects.",
"We propose an approach to learning the semantics of images which allows us to automatically annotate an image with keywords and to retrieve images based on text queries. We do this using a formalism that models the generation of annotated images. We assume that every image is divided into regions, each described by a continuous-valued feature vector. Given a training set of images with annotations, we compute a joint probabilistic model of image features and words which allow us to predict the probability of generating a word given the image regions. This may be used to automatically annotate and retrieve images given a word as a query. Experiments show that our model significantly outperforms the best of the previously reported results on the tasks of automatic image annotation and retrieval.",
"Libraries have traditionally used manual image annotation for indexing and then later retrieving their image collections. However, manual image annotation is an expensive and labor intensive procedure and hence there has been great interest in coming up with automatic ways to retrieve images based on content. Here, we propose an automatic approach to annotating and retrieving images based on a training set of images. We assume that regions in an image can be described using a small vocabulary of blobs. Blobs are generated from image features using clustering. Given a training set of images with annotations, we show that probabilistic models allow us to predict the probability of generating a word given the blobs in an image. This may be used to automatically annotate and retrieve images given a word as a query. We show that relevance models allow us to derive these probabilities in a natural way. Experiments show that the annotation performance of this cross-media relevance model is almost six times as good (in terms of mean precision) than a model based on word-blob co-occurrence model and twice as good as a state of the art model derived from machine translation. Our approach shows the usefulness of using formal information retrieval models for the task of image annotation and retrieval."
]
} |
1705.05102 | 2614174454 | Image search and retrieval engines rely heavily on textual annotation in order to match word queries to a set of candidate images. A system that can automatically annotate images with meaningful text can be highly beneficial for such engines. Currently, the approaches to develop such systems try to establish relationships between keywords and visual features of images. In this paper, We make three main contributions to this area: (i) We transform this problem from the low-level keyword space to the high-level semantics space that we refer to as the " image theme ", (ii) Instead of treating each possible keyword independently, we use latent Dirichlet allocation to learn image themes from the associated texts in a training phase. Images are then annotated with image themes rather than keywords, using a modified continuous relevance model, which takes into account the spatial coherence and the visual continuity among images of common theme. (iii) To achieve more coherent annotations among images of common theme, we have integrated ConceptNet in learning the semantics of images, and hence augment image descriptions beyond annotations provided by humans. Images are thus further annotated by a few most significant words of the prominent image theme. Our extensive experiments show that a coherent theme-based image annotation using high-level semantics results in improved precision and recall as compared with equivalent classical keyword annotation systems. | The annotation problem has been also sometimes treated as a classification problem with class-labels as keywords to be used for annotating images @cite_30 . This approach works well with primitive datasets of very small number of keywords. Some attempts have also been made to incorporate language models and natural language processing tools such as WordNet in the process of image annotation @cite_92 @cite_2 . Some researchers have tried to exploit the correlation between keywords during the process of image annotation, rather than treating each keyword independently of all others @cite_92 @cite_99 @cite_62 . Latent Dirichlet Allocation based image theme modeling was introduced to produce annotation for news images @cite_97 . In this case, each image is accompanied by a news article, which provides additional information regarding that image. worked to establish a similar approach to unify visual and linguistic characteristics of images @cite_55 . conducted a detailed survey of automatic image annotation techniques and arrived at the conclusion that greedy label transfer based approaches can beat complex relevance based algorithms in many cases. They presented two such label transfer based techniques @cite_69 . | {
"cite_N": [
"@cite_30",
"@cite_99",
"@cite_69",
"@cite_62",
"@cite_92",
"@cite_97",
"@cite_55",
"@cite_2"
],
"mid": [
"2888758965",
"",
"1877469910",
"2129758682",
"2138454757",
"1750831471",
"155596317",
"2050113278"
],
"abstract": [
"",
"",
"Automatically assigning keywords to images is of great interest as it allows one to index, retrieve, and understand large collections of image data. Many techniques have been proposed for image annotation in the last decade that give reasonable performance on standard datasets. However, most of these works fail to compare their methods with simple baseline techniques to justify the need for complex models and subsequent training. In this work, we introduce a new baseline technique for image annotation that treats annotation as a retrieval problem. The proposed technique utilizes low-level image features and a simple combination of basic distances to find nearest neighbors of a given image. The keywords are then assigned using a greedy label transfer mechanism. The proposed baseline outperforms the current state-of-the-art methods on two standard and one large Web dataset. We believe that such a baseline measure will provide a strong platform to compare and better understand future annotation techniques.",
"The development of technology generates huge amounts of non-textual information, such as images. An efficient image annotation and retrieval system is highly desired. Clustering algorithms make it possible to represent visual features of images with finite symbols. Based on this, many statistical models, which analyze correspondence between visual features and words and discover hidden semantics, have been published. These models improve the annotation and retrieval of large image databases. However, current state of the art including our previous work produces too many irrelevant keywords for images during annotation. In this paper, we propose a novel approach that augments the classical model with generic knowledge-based, WordNet. Our novel approach strives to prune irrelevant keywords by the usage of WordNet. To identify irrelevant keywords, we investigate various semantic similarity measures between keywords and finally fuse outcomes of all these measures together to make a final decision using Dempster-Shafer evidence combination. We have implemented various models to link visual tokens with keywords based on knowledge-based, WordNet and evaluated performance using precision, and recall using benchmark dataset. The results show that by augmenting knowledge-based with classical model we can improve annotation accuracy by removing irrelevant keywords.",
"Image annotations allow users to access a large image database with textual queries. There have been several studies on automatic image annotation utilizing machine learning techniques, which automatically learn statistical models from annotated images and apply them to generate annotations for unseen images. One common problem shared by most previous learning approaches for automatic image annotation is that each annotated word is predicated for an image independently from other annotated words. In this paper, we proposed a coherent language model for automatic image annotation that takes into account the word-to-word correlation by estimating a coherent language model for an image. This new approach has two important advantages: 1) it is able to automatically determine the annotation length to improve the accuracy of retrieval results, and 2) it can be used with active learning to significantly reduce the required number of annotated image examples. Empirical studies with Corel dataset are presented to show the effectiveness of the coherent language model for automatic image annotation.",
"Image annotation, the task of automatically generating description words for a picture, is a key component in various image search and retrieval applications. Creating image databases for model development is, however, costly and time consuming, since the keywords must be hand-coded and the process repeated for new collections. In this work we exploit the vast resource of images and documents available on the web for developing image annotation models without any human involvement. We describe a probabilistic model based on the assumption that images and their co-occurring textual data are generated by mixtures of latent topics. We show that this model outperforms previously proposed approaches when applied to image annotation and the related task of text illustration despite the noisy nature of our dataset.",
"The question of how meaning might be acquired by young children and represented by adult speakers of a language is one of the most debated topics in cognitive science. Existing semantic representation models are primarily amodal based on information provided by the linguistic input despite ample evidence indicating that the cognitive system is also sensitive to perceptual information. In this work we exploit the vast resource of images and associated documents available on the web and develop a model of multimodal meaning representation which is based on the linguistic and visual context. Experimental results show that a closer correspondence to human data can be obtained by taking the visual modality into account.",
"Automatic image annotation is the task of automatically assigning words to an image that describe the content of the image. Machine learning approaches have been explored to model the association between words and images from an annotated set of images and generate annotations for a test image. The paper proposes methods to use a hierarchy defined on the annotation words derived from a text ontology to improve automatic image annotation and retrieval. Specifically, the hierarchy is used in the context of generating a visual vocabulary for representing images and as a framework for the proposed hierarchical classification approach for automatic image annotation. The effect of using the hierarchy in generating the visual vocabulary is demonstrated by improvements in the annotation performance of translation models. In addition to performance improvements, hierarchical classification approaches yield well to constructing multimedia ontologies."
]
} |
1705.05180 | 2615354843 | Many real-world time-series analysis problems are characterised by scarce data. Solutions typically rely on hand-crafted features extracted from the time or frequency domain allied with classification or regression engines which condition on this (often low-dimensional) feature vector. The huge advances enjoyed by many application domains in recent years have been fuelled by the use of deep learning architectures trained on large data sets. This paper presents an application of deep learning for acoustic event detection in a challenging, data-scarce, real-world problem. Our candidate challenge is to accurately detect the presence of a mosquito from its acoustic signature. We develop convolutional neural networks (CNNs) operating on wavelet transformations of audio recordings. Furthermore, we interrogate the network's predictive power by visualising statistics of network-excitatory samples. These visualisations offer a deep insight into the relative informativeness of components in the detection problem. We include comparisons with conventional classifiers, conditioned on both hand-tuned and generic features, to stress the strength of automatic deep feature learning. Detection is achieved with performance metrics significantly surpassing those of existing algorithmic methods, as well as marginally exceeding those attained by individual human experts. | The process of automatically detecting an acoustic signal in noise typically consists of an initial preprocessing stage, which involves cleaning and denoising the signal itself, followed by a feature extraction process, in which the signal is transformed into a format suitable for a classifier, followed by the final classification stage. Historically, audio feature extraction in signal processing employed domain knowledge and intricate understanding of digital signal theory @cite_17 , leading to hand-crafted feature representations. | {
"cite_N": [
"@cite_17"
],
"mid": [
"2045135321"
],
"abstract": [
"As we look to advance the state of the art in content-based music informatics, there is a general sense that progress is decelerating throughout the field. On closer inspection, performance trajectories across several applications reveal that this is indeed the case, raising some difficult questions for the discipline: why are we slowing down, and what can we do about it? Here, we strive to address both of these concerns. First, we critically review the standard approach to music signal analysis and identify three specific deficiencies to current methods: hand-crafted feature design is sub-optimal and unsustainable, the power of shallow architectures is fundamentally limited, and short-time analysis cannot encode musically meaningful structure. Acknowledging breakthroughs in other perceptual AI domains, we offer that deep learning holds the potential to overcome each of these obstacles. Through conceptual arguments for feature learning and deeper processing architectures, we demonstrate how deep processing models are more powerful extensions of current methods, and why now is the time for this paradigm shift. Finally, we conclude with a discussion of current challenges and the potential impact to further motivate an exploration of this promising research area."
]
} |
1705.04608 | 2951776931 | With the rise of end-to-end learning through deep learning, person detectors and re-identification (ReID) models have recently become very strong. Multi-camera multi-target (MCMT) tracking has not fully gone through this transformation yet. We intend to take another step in this direction by presenting a theoretically principled way of integrating ReID with tracking formulated as an optimal Bayes filter. This conveniently side-steps the need for data-association and opens up a direct path from full images to the core of the tracker. While the results are still sub-par, we believe that this new, tight integration opens many interesting research opportunities and leads the way towards full end-to-end tracking from raw pixels. | Both ReID @cite_36 @cite_24 and tracking @cite_39 come with a vast amount of previous work, for which we refer the interested reader to the cited surveys for a full overview. We will focus on systems that utilize and combine deep learning for both tracking and ReID. | {
"cite_N": [
"@cite_36",
"@cite_24",
"@cite_39"
],
"mid": [
"",
"2531440880",
"1561477459"
],
"abstract": [
"",
"Person re-identification (re-ID) has become increasingly popular in the community due to its application and research significance. It aims at spotting a person of interest in other cameras. In the early days, hand-crafted algorithms and small-scale evaluation were predominantly reported. Recent years have witnessed the emergence of large-scale datasets and deep learning systems which make use of large data volumes. Considering different tasks, we classify most current re-ID methods into two classes, i.e., image-based and video-based; in both tasks, hand-crafted and deep learning systems will be reviewed. Moreover, two new re-ID tasks which are much closer to real-world applications are described and discussed, i.e., end-to-end re-ID and fast re-ID in very large galleries. This paper: 1) introduces the history of person re-ID and its relationship with image classification and instance retrieval; 2) surveys a broad selection of the hand-crafted systems and the large-scale methods in both image- and video-based re-ID; 3) describes critical future directions in end-to-end re-ID and fast retrieval in large galleries; and 4) finally briefs some important yet under-developed issues.",
"Multiple Object Tracking (MOT) is an important computer vision task which has gained increasing attention due to its academic and commercial potential. Although different kinds of approaches have been proposed to tackle this problem, it still has many issues unsolved. For example, factors such as continuous appearance changes and severe occlusions result in difficulties for the task. In order to help the readers understand and learn this topic, we contribute a comprehensive and systematic review. We review the recent advances in various aspects about this topic and propose some interesting directions for future research. To our best knowledge, there has not been any review about this topic in the community. The main contribution of this review is threefold: 1) All key aspects in the multiple object tracking system, including what scenarios the researchers are working on, how their work can be categorized, what needs to be considered when developing a MOT system and how to evaluate a MOT system, are discussed in a clear structure. This review work could not only provide researchers, especially new comers to the topic of MOT, a general understanding of the state-of-the-arts, but also help them to comprehend the aspects of a MOT system and the inter-connected aspects. 2) Instead of listing and summarizing individual publications, we categorize the approaches in the key aspects involved in a MOT system. In each aspect, the methods are divided into different groups and each group is discussed in details for the principles, advances and drawbacks. 3) We provide some potential directions with insights for MOT, which are still open issues and need more research efforts. This would be helpful for researchers to identify further interesting problems."
]
} |
1705.04608 | 2951776931 | With the rise of end-to-end learning through deep learning, person detectors and re-identification (ReID) models have recently become very strong. Multi-camera multi-target (MCMT) tracking has not fully gone through this transformation yet. We intend to take another step in this direction by presenting a theoretically principled way of integrating ReID with tracking formulated as an optimal Bayes filter. This conveniently side-steps the need for data-association and opens up a direct path from full images to the core of the tracker. While the results are still sub-par, we believe that this new, tight integration opens many interesting research opportunities and leads the way towards full end-to-end tracking from raw pixels. | Early work on using the person identity to help tracking is presented by @cite_35 , where classical color-based methods are used. More recent works use CNNs for appearance-based tracking @cite_7 @cite_9 @cite_26 , but typically come with the need of fine-tuning person-specific models online, which might be effective but very costly and is subject to model drift. Leal-Taix ' e al present a siamese network to link similar person boxes @cite_8 , and by this address the complex problem of data association, albeit with the dependency on a detector. Alahi al @cite_19 focus on deep prediction models for multi-target tracking with social long-short term memories (LSTMs). The work of Sadeghian al @cite_33 goes a step further by training LSTMs for ReID, motion, and interaction of persons, but still operates on discrete detector bounding boxes. | {
"cite_N": [
"@cite_35",
"@cite_26",
"@cite_33",
"@cite_7",
"@cite_8",
"@cite_9",
"@cite_19"
],
"mid": [
"2082716591",
"1497265063",
"2951063106",
"823218635",
"2345229200",
"1554825167",
"2424778531"
],
"abstract": [
"We address the problem of multi-person tracking in a complex scene from a single camera. Although tracklet-association methods have shown impressive results in several challenging datasets, discriminability of the appearance model remains a limitation. Inspired by the work of person identity recognition, we obtain discriminative appearance-based affinity models by a novel framework to incorporate the merits of person identity recognition, which help multi-person tracking performance. During off-line learning, a small set of local image descriptors is selected to be used in on-line learned appearances-based affinity models effectively and efficiently. Given short but reliable track-lets generated by frame-to-frame association of detection responses, we identify them as query tracklets and gallery tracklets. For each gallery tracklet, a target-specific appearance model is learned from the on-line training samples collected by spatio-temporal constraints. Both gallery tracklets and query tracklets are fed into hierarchical association framework to obtain final tracking results. We evaluate our proposed system on several public datasets and show significant improvements in terms of tracking evaluation metrics.",
"Convolutional neural network (CNN) models have demonstrated great success in various computer vision tasks including image classification and object detection. However, some equally important tasks such as visual tracking remain relatively unexplored. We believe that a major hurdle that hinders the application of CNN to visual tracking is the lack of properly labeled training data. While existing applications that liberate the power of CNN often need an enormous amount of training data in the order of millions, visual tracking applications typically have only one labeled example in the first frame of each video. We address this research issue here by pre-training a CNN offline and then transferring the rich feature hierarchies learned to online tracking. The CNN is also fine-tuned during online tracking to adapt to the appearance of the tracked target specified in the first video frame. To fit the characteristics of object tracking, we first pre-train the CNN to recognize what is an object, and then propose to generate a probability map instead of producing a simple class label. Using two challenging open benchmarks for performance evaluation, our proposed tracker has demonstrated substantial improvement over other state-of-the-art trackers.",
"We present a multi-cue metric learning framework to tackle the popular yet unsolved Multi-Object Tracking (MOT) problem. One of the key challenges of tracking methods is to effectively compute a similarity score that models multiple cues from the past such as object appearance, motion, or even interactions. This is particularly challenging when objects get occluded or share similar appearance properties with surrounding objects. To address this challenge, we cast the problem as a metric learning task that jointly reasons on multiple cues across time. Our framework learns to encode long-term temporal dependencies across multiple cues with a hierarchical Recurrent Neural Network. We demonstrate the strength of our approach by tracking multiple objects using their appearance, motion, and interactions. Our method outperforms previous works by a large margin on multiple publicly available datasets including the challenging MOT benchmark.",
"Abstract Object appearance model is a crucial module for object tracking and numerous schemes have been developed for object representation with impressive performance. Traditionally, the features used in such object appearance models are predefined in a handcrafted offline way but not tuned for the tracked object. In this paper, we propose a deep learning architecture to learn the most discriminative features dynamically via a convolutional neural network (CNN). In particular, we propose to enhance the discriminative ability of the appearance model in three-fold. First, we design a simple yet effective method to transfer the features learned from CNNs on the source tasks with large scale training data to the new tracking tasks with limited training data. Second, to alleviate the tracker drifting problem caused by model update, we exploit both the ground truth appearance information of the object labeled in the initial frames and the image observations obtained online. Finally, a heuristic schema is used to judge whether updating the object appearance models or not. Extensive experiments on challenging video sequences from the CVPR2013 tracking benchmark validate the robustness and effectiveness of the proposed tracking method.",
"This paper introduces a novel approach to the task of data association within the context of pedestrian tracking, by introducing a two-stage learning scheme to match pairs of detections. First, a Siamese convolutional neural network (CNN) is trained to learn descriptors encoding local spatio-temporal structures between the two input image patches, aggregating pixel values and optical flow information. Second, a set of contextual features derived from the position and size of the compared input patches are combined with the CNN output by means of a gradient boosting classifier to generate the final matching probability. This learning approach is validated by using a linear programming based multi-person tracker showing that even a simple and efficient tracker may outperform much more complex models when fed with our learned matching probabilities. Results on publicly available sequences show that our method meets state-of-the-art standards in multiple people tracking.",
"Deep neural networks, albeit their great success on feature learning in various computer vision tasks, are usually considered as impractical for online visual tracking, because they require very long training time and a large number of training samples. In this paper, we present an efficient and very robust tracking algorithm using a single convolutional neural network (CNN) for learning effective feature representations of the target object in a purely online manner. Our contributions are multifold. First, we introduce a novel truncated structural loss function that maintains as many training samples as possible and reduces the risk of tracking error accumulation. Second, we enhance the ordinary stochastic gradient descent approach in CNN training with a robust sample selection mechanism. The sampling mechanism randomly generates positive and negative samples from different temporal distributions, which are generated by taking the temporal relations and label noise into account. Finally, a lazy yet effective updating scheme is designed for CNN training. Equipped with this novel updating algorithm, the CNN model is robust to some long-existing difficulties in visual tracking, such as occlusion or incorrect detections, without loss of the effective adaption for significant appearance changes. In the experiment, our CNN tracker outperforms all compared state-of-the-art methods on two recently proposed benchmarks, which in total involve over 60 video sequences. The remarkable performance improvement over the existing trackers illustrates the superiority of the feature representations, which are learned purely online via the proposed deep learning framework.",
"Pedestrians follow different trajectories to avoid obstacles and accommodate fellow pedestrians. Any autonomous vehicle navigating such a scene should be able to foresee the future positions of pedestrians and accordingly adjust its path to avoid collisions. This problem of trajectory prediction can be viewed as a sequence generation task, where we are interested in predicting the future trajectory of people based on their past positions. Following the recent success of Recurrent Neural Network (RNN) models for sequence prediction tasks, we propose an LSTM model which can learn general human movement and predict their future trajectories. This is in contrast to traditional approaches which use hand-crafted functions such as Social forces. We demonstrate the performance of our method on several public datasets. Our model outperforms state-of-the-art methods on some of these datasets. We also analyze the trajectories predicted by our model to demonstrate the motion behaviour learned by our model."
]
} |
1705.04608 | 2951776931 | With the rise of end-to-end learning through deep learning, person detectors and re-identification (ReID) models have recently become very strong. Multi-camera multi-target (MCMT) tracking has not fully gone through this transformation yet. We intend to take another step in this direction by presenting a theoretically principled way of integrating ReID with tracking formulated as an optimal Bayes filter. This conveniently side-steps the need for data-association and opens up a direct path from full images to the core of the tracker. While the results are still sub-par, we believe that this new, tight integration opens many interesting research opportunities and leads the way towards full end-to-end tracking from raw pixels. | One of the conceptual differences between the aforementioned works and our explorative work is exactly this dependency on a person detector, providing discrete boxes as starting points. This gives limited state information regarding position, and makes tracking an instance of the complex data association problem. First work towards end-to-end tracking by learning this data association was recently done by @cite_5 . But as mentioned in the introduction, we want to drop both the data association and discrete box representations, and instead keep track of the full belief for each person by leveraging recent ReID models. | {
"cite_N": [
"@cite_5"
],
"mid": [
"2339473870"
],
"abstract": [
"We present a novel approach to online multi-target tracking based on recurrent neural networks (RNNs). Tracking multiple objects in real-world scenes involves many challenges, including a) an a-priori unknown and time-varying number of targets, b) a continuous state estimation of all present targets, and c) a discrete combinatorial problem of data association. Most previous methods involve complex models that require tedious tuning of parameters. Here, we propose for the first time, an end-to-end learning approach for online multi-target tracking. Existing deep learning methods are not designed for the above challenges and cannot be trivially applied to the task. Our solution addresses all of the above points in a principled way. Experiments on both synthetic and real data show promising results obtained at 300 Hz on a standard CPU, and pave the way towards future research in this direction."
]
} |
1705.04601 | 2613853398 | In this paper, we consider the matrices approximated in H2 format. The direct solution, as well as the preconditioning, of systems with such matrices is a challenging problem. We propose a non-extensive sparse factorization of the H2 matrix that allows to substitute the direct H2 solution with the solution of the system with an equivalent sparse matrix of the same size. The sparse factorization is constructed out of parameters of the H2 matrix. In the numerical experiments, we show the consistency of this approach in comparison to the other approximate block low-rank hierarchical solvers, such as HODLR, H2Lib and IFMM. | Hierarchical low-rank matrix formats such as @math royalblue @cite_15 @cite_8 @cite_19 @cite_4 , HODLR @cite_11 @cite_13 royalblue (Hierarchical Off-Diagonal Low-Rank) , HSS @cite_7 @cite_1 @cite_20 royalblue (Hierarchically Semiseparable) , @cite_19 @cite_22 matrices and etc., that are matrix analogies of the fast multipole method @cite_0 @cite_10 , have two significant features: they do store information in data-sparse formats and they provide the fast matrix by vector product. Fast ( @math , where @math is size of the matrix) matrix by vector product allows to apply iterative solvers. Data-sparse representation allows to store matrix in @math cells of memory, but storage scheme is usually complicated. | {
"cite_N": [
"@cite_4",
"@cite_22",
"@cite_7",
"@cite_8",
"@cite_10",
"@cite_1",
"@cite_0",
"@cite_19",
"@cite_15",
"@cite_13",
"@cite_20",
"@cite_11"
],
"mid": [
"2040187556",
"1602760118",
"2009272201",
"1540987201",
"2083206954",
"2141719776",
"",
"",
"2018419001",
"2963754333",
"",
""
],
"abstract": [
"We give a short introduction to methods for the data-sparse approximation of matrices resulting from the discretisation of non-local operators occurring in boundary integral methods, as the inverses of partial differential operators or as solutions of control problems. The result of the approximation will be so-called hierarchical matrices (or short H-matrices). These matrices form a subset of the set of all matrices and have a data-sparse representation. The essential operations for these matrices (matrix-vector and matrix – matrix multiplication, addition and inversion) can be performed in, up to logarithmic factors, optimal complexity. We give a review of specialised variants of H-matrices, especially of H 2 -matrices, and finally consider applications of the different methods to problems from integral equations, partial differential equations and control theory. q 2003 Elsevier Science Ltd. All rights reserved.",
"",
"We describe an algorithm for the direct solution of systems of linear algebraic equations associated with the discretization of boundary integral equations with non-oscillatory kernels in two dimensions. The algorithm is ''fast'' in the sense that its asymptotic complexity is O(n), where n is the number of nodes in the discretization. Unlike previous fast techniques based on iterative solvers, the present algorithm directly constructs a compressed factorization of the inverse of the matrix; thus it is suitable for problems involving relatively ill-conditioned matrices, and is particularly efficient in situations involving multiple right hand sides. The performance of the scheme is illustrated with several numerical examples. rformance of the scheme is illustrated with several numerical examples. ples.",
"The preceding Part I of this paper has introduced a class of matrices (ℋ-matrices) which are data-sparse and allow an approximate matrix arithmetic of almost linear complexity. The matrices discussed in Part I are able to approximate discrete integral operators in the case of one spatial dimension.",
"An algorithm is presented for the rapid evaluation of the potential and force fields in systems involving large numbers of particles whose interactions are Coulombic or gravitational in nature. For a system ofNparticles, an amount of work of the orderO(N2) has traditionally been required to evaluate all pairwise interactions, unless some approximation or truncation method is used. The algorithm of the present paper requires an amount of work proportional toNto evaluate all interactions to within roundoff error, making it considerably more practical for large-scale problems encountered in plasma physics, fluid dynamics, molecular dynamics, and celestial mechanics.",
"In this paper we present a fast direct solver for certain classes of dense structured linear systems that works by first converting the given dense system to a larger system of block sparse equations and then uses standard sparse direct solvers. The kind of matrix structures that we consider are induced by numerical low rank in the off-diagonal blocks of the matrix and are related to the structures exploited by the fast multipole method (FMM) of Greengard and Rokhlin. The special structure that we exploit in this paper is captured by what we term the hierarchically semiseparable (HSS) representation of a matrix. Numerical experiments indicate that the method is probably backward stable.",
"",
"",
"A class of matrices ( ( H )-matrices) is introduced which have the following properties. (i) They are sparse in the sense that only few data are needed for their representation. (ii) The matrix-vector multiplication is of almost linear complexity. (iii) In general, sums and products of these matrices are no longer in the same set, but their truncations to the ( H )-matrix format are again of almost linear complexity. (iv) The same statement holds for the inverse of an ( H )-matrix.",
"A number of problems in probability and statistics can be addressed using the multivariate normal (Gaussian) distribution. In the one-dimensional case, computing the probability for a given mean and variance simply requires the evaluation of the corresponding Gaussian density. In the @math -dimensional setting, however, it requires the inversion of an @math covariance matrix, @math , as well as the evaluation of its determinant, @math . In many cases, such as regression using Gaussian processes, the covariance matrix is of the form @math , where @math is computed using a specified covariance kernel which depends on the data and additional parameters (hyperparameters). The matrix @math is typically dense, causing standard direct methods for inversion and determinant evaluation to require @math work. This cost is prohibitive for large-scale modeling. Here, we show that for the most commonly used covariance functions, the matrix @math can be hierarchically factored into a product of block low-rank updates of the identity matrix, yielding an @math algorithm for inversion. More importantly, we show that this factorization enables the evaluation of the determinant @math , permitting the direct calculation of probabilities in high dimensions under fairly broad assumptions on the kernel defining @math . Our fast algorithm brings many problems in marginalization and the adaptation of hyperparameters within practical reach using a single CPU core. The combination of nearly optimal scaling in terms of problem size with high-performance computing resources will permit the modeling of previously intractable problems. We illustrate the performance of the scheme on standard covariance kernels.",
"",
""
]
} |
1705.04601 | 2613853398 | In this paper, we consider the matrices approximated in H2 format. The direct solution, as well as the preconditioning, of systems with such matrices is a challenging problem. We propose a non-extensive sparse factorization of the H2 matrix that allows to substitute the direct H2 solution with the solution of the system with an equivalent sparse matrix of the same size. The sparse factorization is constructed out of parameters of the H2 matrix. In the numerical experiments, we show the consistency of this approach in comparison to the other approximate block low-rank hierarchical solvers, such as HODLR, H2Lib and IFMM. | The factorization approach is more popular for hierarchical matrices with strong low-rank structure, also known as hierarchical matrices with weak-admissibility criteria royalblue @cite_25 ( @math royalblue @cite_15 @cite_8 , HODLR royalblue @cite_11 @cite_13 , HSS royalblue @cite_7 @cite_1 @cite_20 matrices). royalblue For the @math matrix, the algorithm @math -LU @cite_6 with almost linear complexity was proposed. This algorithm has been successfully applied to many problems. The major drawback of the @math -LU algorithm is that factorization time and memory required for @math and @math factors can be quite large. Approximate direct solvers based on factorization of HSS and HODLR matrices are royalblue also well studied royalblue and found many @cite_20 @cite_17 @cite_21 @cite_3 @cite_9 @cite_13 @cite_7 successful applications . | {
"cite_N": [
"@cite_7",
"@cite_8",
"@cite_9",
"@cite_21",
"@cite_1",
"@cite_17",
"@cite_6",
"@cite_3",
"@cite_15",
"@cite_13",
"@cite_25",
"@cite_20",
"@cite_11"
],
"mid": [
"2009272201",
"1540987201",
"",
"1983103429",
"2141719776",
"2075490698",
"2088527939",
"1252747413",
"2018419001",
"2963754333",
"1976016599",
"",
""
],
"abstract": [
"We describe an algorithm for the direct solution of systems of linear algebraic equations associated with the discretization of boundary integral equations with non-oscillatory kernels in two dimensions. The algorithm is ''fast'' in the sense that its asymptotic complexity is O(n), where n is the number of nodes in the discretization. Unlike previous fast techniques based on iterative solvers, the present algorithm directly constructs a compressed factorization of the inverse of the matrix; thus it is suitable for problems involving relatively ill-conditioned matrices, and is particularly efficient in situations involving multiple right hand sides. The performance of the scheme is illustrated with several numerical examples. rformance of the scheme is illustrated with several numerical examples. ples.",
"The preceding Part I of this paper has introduced a class of matrices (ℋ-matrices) which are data-sparse and allow an approximate matrix arithmetic of almost linear complexity. The matrices discussed in Part I are able to approximate discrete integral operators in the case of one spatial dimension.",
"",
"Randomized sampling has recently been proven a highly efficient technique for computing approximate factorizations of matrices that have low numerical rank. This paper describes an extension of such techniques to a wider class of matrices that are not themselves rank-deficient but have off-diagonal blocks that are; specifically, the class of so-called hierarchically semiseparable (HSS) matrices. HSS matrices arise frequently in numerical analysis and signal processing, particularly in the construction of fast methods for solving differential and integral equations numerically. The HSS structure admits algebraic operations (matrix-vector multiplications, matrix factorizations, matrix inversion, etc.) to be performed very rapidly, but only once the HSS representation of the matrix has been constructed. How to rapidly compute this representation in the first place is much less well understood. The present paper demonstrates that if an @math matrix can be applied to a vector in @math time, and if individual entries of the matrix can be computed rapidly, then provided that an HSS representation of the matrix exists, it can be constructed in @math operations, where @math is an upper bound for the numerical rank of the off-diagonal blocks. The point is that when legacy codes (based on, e.g., the fast multipole method) can be used for the fast matrix-vector multiply, the proposed algorithm can be used to obtain the HSS representation of the matrix, and then well-established techniques for HSS matrices can be used to invert or factor the matrix.",
"In this paper we present a fast direct solver for certain classes of dense structured linear systems that works by first converting the given dense system to a larger system of block sparse equations and then uses standard sparse direct solvers. The kind of matrix structures that we consider are induced by numerical low rank in the off-diagonal blocks of the matrix and are related to the structures exploited by the fast multipole method (FMM) of Greengard and Rokhlin. The special structure that we exploit in this paper is captured by what we term the hierarchically semiseparable (HSS) representation of a matrix. Numerical experiments indicate that the method is probably backward stable.",
"In this paper we develop a fast direct solver for large discretized linear systems using the supernodal multifrontal method together with low-rank approximations. For linear systems arising from certain partial differential equations such as elliptic equations, during the Gaussian elimination of the matrices with proper ordering, the fill-in has a low-rank property: all off-diagonal blocks have small numerical ranks with proper definition of off-diagonal blocks. Matrices with this low-rank property can be efficiently approximated with semiseparable structures called hierarchically semiseparable (HSS) representations. We reveal the above low-rank property by ordering the variables with nested dissection and eliminating them with the multifrontal method. All matrix operations in the multifrontal method are performed in HSS forms. We present efficient ways to organize the HSS structured operations along the elimination. Some fast HSS matrix operations using tree structures are proposed. This new structured multifrontal method has nearly linear complexity and a linear storage requirement. Thus, we call it a superfast multifrontal method. It is especially suitable for large sparse problems and also has natural adaptability to parallel computations and great potential to provide effective preconditioners. Numerical results demonstrate the efficiency.",
"The adaptive cross approximation method can be used to efficiently approximate stiffness matrices arising from boundary element applications by hierarchical matrices. In this article an approximative LU decomposition in the same format is presented which can be used for preconditioning the resulting coefficient matrices efficiently. If the LU decomposition is computed with high precision, it may even be used as a direct yet efficient solver.",
"This paper presents a fast direct solver for 3D discretized linear systems using the supernodal multifrontal method together with low-rank approximations. For linear systems arising from certain partial differential equations (PDEs) such as elliptic equations, during the Gaussian elimination of the matrices with Nested Dissection ordering, the fill-in of L and U factors loses its sparsity and contains dense blocks with low-rank property. Off-diagonal blocks can be efficiently approximated with low-rank matrices; diagonal blocks approximated with semiseparable structures called hierarchically semiseparable (HSS) representations. Matrix operations in the multifrontal method are performed in low-rank arithmetic. We present efficient way to organize the HSS structured operations along the elimination. To compress dense blocks into low-rank or HSS structures, we use effective cross approximation (CA) approach. We also use idea of adaptive balancing between robust arithmetic for computing the small dense blocks and low-rank matrix operations for handling with compressed ones while performing the Gaussian elimination. This new proposed solver can be essentially parallelized both on architecture with shared and distributed memory and can be used as effective preconditioner. To check efficient of our solver we compare it with Intel MKL PARDISO - the high performance direct solver. Memory and performance tests demonstrate up to 3 times performance and memory gain for the 3D problems with more than (10^6 ) unknowns. Therefore, proposed multifrontal HSS solver can solve large problems, which cannot be resolved by direct solvers because of large memory consumptions.",
"A class of matrices ( ( H )-matrices) is introduced which have the following properties. (i) They are sparse in the sense that only few data are needed for their representation. (ii) The matrix-vector multiplication is of almost linear complexity. (iii) In general, sums and products of these matrices are no longer in the same set, but their truncations to the ( H )-matrix format are again of almost linear complexity. (iv) The same statement holds for the inverse of an ( H )-matrix.",
"A number of problems in probability and statistics can be addressed using the multivariate normal (Gaussian) distribution. In the one-dimensional case, computing the probability for a given mean and variance simply requires the evaluation of the corresponding Gaussian density. In the @math -dimensional setting, however, it requires the inversion of an @math covariance matrix, @math , as well as the evaluation of its determinant, @math . In many cases, such as regression using Gaussian processes, the covariance matrix is of the form @math , where @math is computed using a specified covariance kernel which depends on the data and additional parameters (hyperparameters). The matrix @math is typically dense, causing standard direct methods for inversion and determinant evaluation to require @math work. This cost is prohibitive for large-scale modeling. Here, we show that for the most commonly used covariance functions, the matrix @math can be hierarchically factored into a product of block low-rank updates of the identity matrix, yielding an @math algorithm for inversion. More importantly, we show that this factorization enables the evaluation of the determinant @math , permitting the direct calculation of probabilities in high dimensions under fairly broad assumptions on the kernel defining @math . Our fast algorithm brings many problems in marginalization and the adaptation of hyperparameters within practical reach using a single CPU core. The combination of nearly optimal scaling in terms of problem size with high-performance computing resources will permit the modeling of previously intractable problems. We illustrate the performance of the scheme on standard covariance kernels.",
"In preceding papers [8], [11], [12], [6], a class of matrices ( Open image in new window -matrices) has been developed which are data-sparse and allow to approximate integral and more general nonlocal operators with almost linear complexity. In the present paper, a weaker admissibility condition is described which leads to a coarser partitioning of the hierarchical Open image in new window -matrix format. A coarser format yields smaller constants in the work and storage estimates and thus leads to a lower complexity of the Open image in new window -matrix arithmetic. On the other hand, it preserves the approximation power which is known in the case of the standard admissibility criterion. Furthermore, the new weak Open image in new window -matrix format allows to analyse the accuracy of the Open image in new window -matrix inversion and multiplication.",
"",
""
]
} |
1705.04138 | 2613172232 | We consider the stochastic composition optimization problem proposed in wang2017stochastic , which has applications ranging from estimation to statistical and machine learning. We propose the first ADMM-based algorithm named com-SVR-ADMM, and show that com-SVR-ADMM converges linearly for strongly convex and Lipschitz smooth objectives, and has a convergence rate of @math , which improves upon the @math rate in wang2016accelerating when the objective is convex and Lipschitz smooth. Moreover, com-SVR-ADMM possesses a rate of @math when the objective is convex but without Lipschitz smoothness. We also conduct experiments and show that it outperforms existing algorithms. | The stochastic composition optimization problem was first proposed in @cite_13 , where two solution algorithms, Basic SCGD and accelerated SCGD, were proposed. The algorithms were shown to achieve a sublinear convergence rate for convex and strongly convex cases, and were also shown to converge to a stationary point in the nonconvex case. Later, @cite_0 proposed a proximal gradient algorithm called ASC-PG to improve the convergence rate when both inner and outer functions are smooth. However, the convergence rate is sublinear and their results do not include the regularizer when the objective functions are not strongly convex. In @cite_4 , the authors solved the finite sample case of stochastic composition optimization and obtained two linear-convergent algorithms based on the stochastic variance reduction gradient technique (SVRG) proposed in @cite_15 . However, the algorithms do not handle the regularizer either. | {
"cite_N": [
"@cite_0",
"@cite_15",
"@cite_13",
"@cite_4"
],
"mid": [
"2512803856",
"",
"1494085563",
"2538933862"
],
"abstract": [
"Consider the stochastic composition optimization problem where the objective is a composition of two expected-value functions. We propose a new stochastic first-order method, namely the accelerated stochastic compositional proximal gradient (ASC-PG) method, which updates based on queries to the sampling oracle using two different timescales. The ASC-PG is the first proximal gradient method for the stochastic composition problem that can deal with nonsmooth regularization penalty. We show that the ASC-PG exhibits faster convergence than the best known algorithms, and that it achieves the optimal sample-error complexity in several important special cases. We further demonstrate the application of ASC-PG to reinforcement learning and conduct numerical experiments.",
"",
"Classical stochastic gradient methods are well suited for minimizing expected-value objective functions. However, they do not apply to the minimization of a nonlinear function involving expected values or a composition of two expected-value functions, i.e., the problem @math minxEvfv(Ew[gw(x)]). In order to solve this stochastic composition problem, we propose a class of stochastic compositional gradient descent (SCGD) algorithms that can be viewed as stochastic versions of quasi-gradient method. SCGD update the solutions based on noisy sample gradients of @math fv,gw and use an auxiliary variable to track the unknown quantity @math Ewgw(x). We prove that the SCGD converge almost surely to an optimal solution for convex optimization problems, as long as such a solution exists. The convergence involves the interplay of two iterations with different time scales. For nonsmooth convex problems, the SCGD achieves a convergence rate of @math O(k-1 4) in the general case and @math O(k-2 3) in the strongly convex case, after taking k samples. For smooth convex problems, the SCGD can be accelerated to converge at a rate of @math O(k-2 7) in the general case and @math O(k-4 5) in the strongly convex case. For nonconvex problems, we prove that any limit point generated by SCGD is a stationary point, for which we also provide the convergence rate analysis. Indeed, the stochastic setting where one wants to optimize compositions of expected-value functions is very common in practice. The proposed SCGD methods find wide applications in learning, estimation, dynamic programming, etc.",
"The stochastic composition optimization proposed recently by [2014] minimizes the objective with the compositional expectation form: @math It summarizes many important applications in machine learning, statistics, and finance. In this paper, we consider the finite-sum scenario for composition optimization: [ f (x) := 1 n i = 1 ^n F_i ( 1 m j = 1 ^m G_j (x) ). ] We propose two algorithms to solve this problem by combining the stochastic compositional gradient descent (SCGD) and the stochastic variance reduced gradient (SVRG) technique. A constant linear convergence rate is proved for strongly convex optimization, which substantially improves the sublinear rate @math of the best known algorithm."
]
} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.