aid stringlengths 9 15 | mid stringlengths 7 10 | abstract stringlengths 78 2.56k | related_work stringlengths 92 1.77k | ref_abstract dict |
|---|---|---|---|---|
1907.10599 | 2963790895 | Are neural networks biased toward simple functions? Does depth always help learn more complex features? Is training the last layer of a network as good as training all layers? These questions seem unrelated at face value, but in this work we give all of them a common treatment from the spectral perspective. We will stu... | @cite_2 presents a common framework, known as , unifying the GP, NTK, signal propagation, and random matrix perspectives, as well as extending them to new scenarios, like recurrent neural networks. It proves the existence of and allows the computation of a large number of infinite-width limits (including ones relevant ... | {
"cite_N": [
"@cite_2"
],
"mid": [
"2913473169"
],
"abstract": [
"Several recent trends in machine learning theory and practice, from the design of state-of-the-art Gaussian Process to the convergence analysis of deep neural nets (DNNs) under stochastic gradient descent (SGD), have found it fruit... |
1907.10594 | 2964263576 | The health effects of air pollution have been subject to intense study in recent decades. Exposure to pollutants such as airborne particulate matter and ozone has been associated with increases in morbidity and mortality, especially with regards to respiratory and cardiovascular diseases. Unfortunately, individuals do ... | Breathing rate and tidal volume are estimated from heart rate during exercise. The breathing rate and tidal volume vary in response to metabolic demand and increase in physical activity @cite_5 .First, the tidal volume of the individual is calculated which is the lung volume representing the normal volume of air displa... | {
"cite_N": [
"@cite_5",
"@cite_1",
"@cite_4"
],
"mid": [
"2069615521",
"2597070792",
"2011652354"
],
"abstract": [
"Background: The power of breathing (PoB) is used to estimate the mechanical workload of the respiratory system. Aim of this study was to investigate the effect of di... |
1907.10594 | 2964263576 | The health effects of air pollution have been subject to intense study in recent decades. Exposure to pollutants such as airborne particulate matter and ozone has been associated with increases in morbidity and mortality, especially with regards to respiratory and cardiovascular diseases. Unfortunately, individuals do ... | The cigarette equivalent is derived from both the pollution intake and the tidal volume, where depending on the percentage of each of the pollutant intake, the total summation gives the result. Here the cigarette equivalent is used to eliminate the confusion, by converting real-time air quality data from pollution leve... | {
"cite_N": [
"@cite_8",
"@cite_11"
],
"mid": [
"2343449564",
"2135548528"
],
"abstract": [
"Abstract Background Although the health effects of long term exposure to air pollution are well established, it is difficult to effectively communicate the health risks of this (largely invisible) ... |
1907.10416 | 2963631953 | This paper proposes a method to guide tensor factorization, using class labels. Furthermore, it shows the advantages of using the proposed method in identifying nodes that play a special role in multi-relational networks, e.g. spammers. Most complex systems involve multiple types of relationships and interactions among... | A recent approach to tensor factorization is RESCAL @cite_16 which achieves high predictive performance in the task of link prediction. RESCAL, which we will describe in more detail in Section , uses a unique latent representation for entities. | {
"cite_N": [
"@cite_16"
],
"mid": [
"2045716015"
],
"abstract": [
"The constant advances in sequencing technology have redefined the way genome sequencing is performed. They are able to produce tens of millions of short sequences (reads), during a single experiment, and with a much lower cost tha... |
1907.10416 | 2963631953 | This paper proposes a method to guide tensor factorization, using class labels. Furthermore, it shows the advantages of using the proposed method in identifying nodes that play a special role in multi-relational networks, e.g. spammers. Most complex systems involve multiple types of relationships and interactions among... | As multi-relational data can be efficiently represented by tensors, TripleRank @cite_8 employs tensor factorization in order to rank entities in the context of linked data. TripleRank applies a common tensor factorization CANDECOMP PARAFAC @cite_14 to obtain two factor matrices which correspond to hub and authority sco... | {
"cite_N": [
"@cite_14",
"@cite_3",
"@cite_8"
],
"mid": [
"1974403130",
"2056088289",
"1910578190"
],
"abstract": [
"We review the method of Parallel Factor Analysis, which simultaneously fits multiple two-way arrays or ‘slices’ of a three-way array in terms of a common set of fac... |
1907.10360 | 2963382051 | We consider multi-agent transport task problems where, e.g. in a factory setting, items have to be delivered from a given start to a goal pose while the delivering robots need to avoid collisions with each other on the floor. We introduce a Task Conflict-Based Search (TCBS) Algorithm to solve the combined delivery task... | The particular problem of transport task allocation is called PDP @cite_10 in operations research. See @cite_16 for a survey article. It is defined by a set of agents and a number of requests of certain amount to be transported from one location to another. The problem is often studied with time windows @cite_14 or to ... | {
"cite_N": [
"@cite_14",
"@cite_5",
"@cite_16",
"@cite_10",
"@cite_12"
],
"mid": [
"2212025445",
"2106601208",
"1525106674",
"2063659595",
"2005974211"
],
"abstract": [
"In the pickup and delivery problem with time windows (PDPTW), vehicles have to transport loads ... |
1907.10360 | 2963382051 | We consider multi-agent transport task problems where, e.g. in a factory setting, items have to be delivered from a given start to a goal pose while the delivering robots need to avoid collisions with each other on the floor. We introduce a Task Conflict-Based Search (TCBS) Algorithm to solve the combined delivery task... | MAPF is another intensively studied problem in multi-agent systems @cite_33 . The decision in MAPF concerns how a number of agents will be traveling to their goal poses without colliding, also an NP-hard problem @cite_15 . The colored pebble problem is comparable since the color of the pebble makes them not interchange... | {
"cite_N": [
"@cite_30",
"@cite_31",
"@cite_33",
"@cite_7",
"@cite_22",
"@cite_29",
"@cite_32",
"@cite_19",
"@cite_15",
"@cite_13",
"@cite_11"
],
"mid": [
"",
"2016374168",
"",
"2163502413",
"2621058035",
"",
"2963128936",
"2292985786",
... |
1907.10360 | 2963382051 | We consider multi-agent transport task problems where, e.g. in a factory setting, items have to be delivered from a given start to a goal pose while the delivering robots need to avoid collisions with each other on the floor. We introduce a Task Conflict-Based Search (TCBS) Algorithm to solve the combined delivery task... | One problem formulation that is more closely related to transport systems is the TAPF introduced by Ma and Koenig @cite_23 . It first solves the assignment problem and MAPF problem but not concurrently, so the costs used for task allocation are not the true costs. Instead of single goals we consider the whole transport... | {
"cite_N": [
"@cite_23"
],
"mid": [
"2482025661"
],
"abstract": [
"We study the TAPF (combined target-assignment and path-finding) problem for teams of agents in known terrain, which generalizes both the anonymous and non-anonymous multi-agent path-finding problems. Each of the teams is given the... |
1907.10360 | 2963382051 | We consider multi-agent transport task problems where, e.g. in a factory setting, items have to be delivered from a given start to a goal pose while the delivering robots need to avoid collisions with each other on the floor. We introduce a Task Conflict-Based Search (TCBS) Algorithm to solve the combined delivery task... | A different type of problem is studied in vehicle routing with capacities @cite_4 , which focuses on the deliveries from one central depot based on certain demands. This is a different problem in the sense that it considers only one origin and additionally considers capacity constraints. | {
"cite_N": [
"@cite_4"
],
"mid": [
"2050205301"
],
"abstract": [
"Genetic Algorithms (GAs) can efficiently produce high quality results for hard combinatorial real world problems such as the Vehicle Routing Problem (VRP). Genetic Vehicle Representation (GVR), a recent approach to solving instance... |
1907.10360 | 2963382051 | We consider multi-agent transport task problems where, e.g. in a factory setting, items have to be delivered from a given start to a goal pose while the delivering robots need to avoid collisions with each other on the floor. We introduce a Task Conflict-Based Search (TCBS) Algorithm to solve the combined delivery task... | The joint solution of MATA and MAPF , that we are proposing, was previously studied in @cite_9 , where the problem is considered as MILP but due to collisions on the single-agent path level we think it should be considered a MINLP Problem. Therefore, the solver proposed by @cite_9 can not solve the problem optimally si... | {
"cite_N": [
"@cite_9"
],
"mid": [
"42384017"
],
"abstract": [
"Existing approaches to multirobot coordination separate scheduling and task allocation, but finding the optimal schedule with joint tasks and spatial constraints requires robots to simultaneously solve the scheduling, task allocation... |
1907.10360 | 2963382051 | We consider multi-agent transport task problems where, e.g. in a factory setting, items have to be delivered from a given start to a goal pose while the delivering robots need to avoid collisions with each other on the floor. We introduce a Task Conflict-Based Search (TCBS) Algorithm to solve the combined delivery task... | A similar problem, by taking uncertainties into account, aims at applications in highly dynamic environments @cite_2 . The solution is also sub-optimal because agent-agent collisions are not considered at planning time. | {
"cite_N": [
"@cite_2"
],
"mid": [
"2395113136"
],
"abstract": [
"We consider the multi-robot task allocation (MRTA) problem in an initially unknown environment. The objective of the MRTA problem is to find a schedule or sequence of tasks that should be performed by a set of robots so that the co... |
1907.10360 | 2963382051 | We consider multi-agent transport task problems where, e.g. in a factory setting, items have to be delivered from a given start to a goal pose while the delivering robots need to avoid collisions with each other on the floor. We introduce a Task Conflict-Based Search (TCBS) Algorithm to solve the combined delivery task... | Similar problems have recently also been formulated by @cite_20 , @cite_18 and @cite_8 for domains that do not involve transport tasks as we consider them. Both find interesting sub-optimal solutions for TPTS @cite_18 , based on DCOP @cite_20 and using answer set programming (ASP) @cite_8 . All of these approaches may ... | {
"cite_N": [
"@cite_18",
"@cite_20",
"@cite_8"
],
"mid": [
"2962969317",
"2515724774",
"2741892665"
],
"abstract": [
"The multi-agent path-finding (MAPF) problem has recently received a lot of attention. However, it does not capture important characteristics of many real-world dom... |
1907.10360 | 2963382051 | We consider multi-agent transport task problems where, e.g. in a factory setting, items have to be delivered from a given start to a goal pose while the delivering robots need to avoid collisions with each other on the floor. We introduce a Task Conflict-Based Search (TCBS) Algorithm to solve the combined delivery task... | The field of @cite_27 is also combining two planning domains but for mobile manipulators. Our algorithm borrows the concept of hierarchical planners and the reaction to planning errors i.e. when no path is found. | {
"cite_N": [
"@cite_27"
],
"mid": [
"2143936212"
],
"abstract": [
"We present a hierarchical planning system and its application to robotic manipulation. The novel features of the system are: 1) it finds high-quality kinematic solutions to task-level problems; 2) it takes advantage of subtask-spe... |
1907.10371 | 2963170514 | Comments on social media are very diverse, in terms of content, style and vocabulary, which make generating comments much more challenging than other existing natural language generation (NLG) tasks. Besides, since different user has different expression habits, it is necessary to take the user's profile into considera... | This paper focuses on comments generation task, which can be further divided into generating a comment according to the structure data @cite_1 , text data @cite_7 , image @cite_0 and video @cite_4 , separately. | {
"cite_N": [
"@cite_0",
"@cite_4",
"@cite_1",
"@cite_7"
],
"mid": [
"1895577753",
"2890638727",
"2204900930",
"2963280377"
],
"abstract": [
"Automatically describing the content of an image is a fundamental problem in artificial intelligence that connects computer vision a... |
1907.10451 | 2963188742 | How to perform effective information fusion of different modalities is a core factor in boosting the performance of RGBT tracking. This paper presents a novel deep fusion algorithm based on the representations from an end-to-end trained convolutional neural network. To deploy the complementarity of features of all laye... | RGBT tracking receives more and more attention in the computer vision community with the popularity of thermal infrared sensors. Recent methods on RGBT tracking mainly focus on sparse representation because of its capability of suppressing noise and errors @cite_35 , @cite_15 . Wu @cite_35 concatenate the image patches... | {
"cite_N": [
"@cite_35",
"@cite_15",
"@cite_37"
],
"mid": [
"2106944077",
"",
"2896228140"
],
"abstract": [
"Information from multiple heterogenous data sources (e.g. visible and infrared) or representations (e.g. intensity and edge) have become increasingly important in many vide... |
1907.10451 | 2963188742 | How to perform effective information fusion of different modalities is a core factor in boosting the performance of RGBT tracking. This paper presents a novel deep fusion algorithm based on the representations from an end-to-end trained convolutional neural network. To deploy the complementarity of features of all laye... | Feature aggregation @cite_30 @cite_23 is becoming more and more popular to improve network performance by enhancing the representation of features. Without exception, in the field of visual tracking @cite_16 @cite_33 @cite_2 @cite_50 , there are many methods to improve tracking performance by the skill of feature aggre... | {
"cite_N": [
"@cite_30",
"@cite_26",
"@cite_33",
"@cite_23",
"@cite_2",
"@cite_50",
"@cite_16"
],
"mid": [
"2738535327",
"2963534981",
"2473868734",
"",
"2557641257",
"2518013266",
"2775609985"
],
"abstract": [
"Visual recognition requires rich repr... |
1907.10484 | 2966674341 | Abstract Audit logs serve as a critical component in enterprise business systems and are used for auditing, storing, and tracking changes made to the data. However, audit logs are vulnerable to a series of attacks enabling adversaries to tamper data and corresponding audit logs without getting detected. Among them, two... | Blockchain and Audit Logs Combining blockchain and audit logs, Sutton and Samvi @cite_52 proposed a blockchain-based approach that stores the integrity proof digest to the Bitcoin blockchain. Bitcoin uses a proof-of-work (PoW) consensus protocol. As we show later in tab:ca , PoW suffers from low throughput and high con... | {
"cite_N": [
"@cite_52",
"@cite_7",
"@cite_56"
],
"mid": [
"2763836263",
"2520906649",
"2883678950"
],
"abstract": [
"Privacy audit logs are used to capture the actions of participants in a data sharing environment in order for auditors to check compliance with privacy policies. H... |
1907.10588 | 2966220704 | Crowdsourcing platforms enable companies to propose tasks to a large crowd of users. The workers receive a compensation for their work according to the serious of the tasks they managed to accomplish. The evaluation of the quality of responses obtained from the crowd remains one of the most important problems in this c... | The identification of experts on crowdsourcing platforms has been the subject of several recent studies. Two different types of approach have been used: the ones where no prior knowledge is available and the ones using questions whose correct answers are known in advance. These questions with their known values are cal... | {
"cite_N": [
"@cite_0",
"@cite_7"
],
"mid": [
"2515572806",
"1993853885"
],
"abstract": [
"Crowdsourcing platforms enable to propose simple human intelligence tasks to a large number of participants who realise these tasks. The workers often receive a small amount of money or the platform... |
1907.10588 | 2966220704 | Crowdsourcing platforms enable companies to propose tasks to a large crowd of users. The workers receive a compensation for their work according to the serious of the tasks they managed to accomplish. The evaluation of the quality of responses obtained from the crowd remains one of the most important problems in this c... | @cite_9 and @cite_1 also used this approach for binary classifications and categorical labeling. @cite_13 have generalized this technique on ordinary rankings (associating scores from 1 to 5 depending on the quality of an object or a service). These methods converge to calculate the sensitivity'' (the true positives) a... | {
"cite_N": [
"@cite_13",
"@cite_9",
"@cite_1",
"@cite_2"
],
"mid": [
"2119830539",
"2144660879",
"2134305421",
"2187291759"
],
"abstract": [
"With the advent of crowdsourcing services it has become quite cheap and reasonably effective to get a data set labeled by multiple ... |
1907.10468 | 2962887558 | We revisit the complexity of deciding, given a bimatrix game, whether it has a Nash equilibrium with certain natural properties; such decision problems were early known to be @math -hard GZ89 . We show that @math -hardness still holds under two significant restrictions in simultaneity: the game is win-lose (that is, al... | None of the works @cite_21 @cite_30 @cite_13 @cite_5 @cite_3 @cite_18 @cite_9 @cite_19 on the complexity of decision and counting problems about Nash equilibria in bimatrix games considered the two restrictions to win-lose bimatrix and symmetric bimatrix games in simultaneity; neither did the works @cite_0 @cite_7 @cit... | {
"cite_N": [
"@cite_30",
"@cite_18",
"@cite_7",
"@cite_9",
"@cite_21",
"@cite_3",
"@cite_0",
"@cite_19",
"@cite_27",
"@cite_23",
"@cite_5",
"@cite_13"
],
"mid": [
"2149624304",
"2091731906",
"2057913812",
"2104647222",
"2104892205",
"2021939... |
1907.10398 | 2963225856 | The median of a graph @math is the set of all vertices @math of @math minimizing the sum of distances from @math to all other vertices of @math . It is known that computing the median of dense graphs in subcubic time refutes the APSP conjecture and computing the median of sparse graphs in subquadratic time refutes the ... | As noticed above, the @math -classes of a median graph @math correspond to coordinates of the hypercube in which @math isometrically embeds. Thus one can define @math -classes for all partial cubes. Eppstein @cite_29 performed an efficient computation of @math -classes as a main step of his @math algorithm for recogniz... | {
"cite_N": [
"@cite_22",
"@cite_9",
"@cite_29",
"@cite_3",
"@cite_49"
],
"mid": [
"2073943982",
"1545078593",
"2016084255",
"2094731111",
""
],
"abstract": [
"Abstract Motivated by a dynamic location problem for graphs, Chung, Graham and Saks introduced a graph par... |
1907.10406 | 2963805462 | Deep neural networks are becoming popular and important assets of many AI companies. However, recent studies indicate that they are also vulnerable to adversarial attacks. Adversarial attacks can be either white-box or black-box. The white-box attacks assume full knowledge of the models while the black-box ones assume ... | In this work, we aim to identify the internal DNN architecture @cite_6 @cite_7 . In the real-world applications, several DNN architectures are widely used. For example, AlexNet @cite_0 is popular for its success in the 2012 ImageNet competition @cite_35 . GoogleNet @cite_9 significantly increases the depth of DNN. ResN... | {
"cite_N": [
"@cite_30",
"@cite_35",
"@cite_18",
"@cite_14",
"@cite_7",
"@cite_9",
"@cite_6",
"@cite_0",
"@cite_23",
"@cite_31"
],
"mid": [
"2783000019",
"2117539524",
"2612445135",
"2194775991",
"",
"2097117768",
"2072128103",
"2163605009",... |
1907.10406 | 2963805462 | Deep neural networks are becoming popular and important assets of many AI companies. However, recent studies indicate that they are also vulnerable to adversarial attacks. Adversarial attacks can be either white-box or black-box. The white-box attacks assume full knowledge of the models while the black-box ones assume ... | Side-channel attack (SCA) is a very powerful tool in attacking encrypted systems. Traditionally, the encryption process is considered as a perfect black-box. However, in real-world applications, information can be leaking @cite_38 . Initially, SCA is focused on differential power analysis @cite_4 and timing attacks @ci... | {
"cite_N": [
"@cite_38",
"@cite_37",
"@cite_4",
"@cite_33",
"@cite_8",
"@cite_15",
"@cite_17"
],
"mid": [
"2402794349",
"1664808737",
"2154909745",
"1427174644",
"2108742901",
"2294912729",
""
],
"abstract": [
"Side-channel attacks are easy-to-imple... |
1907.10406 | 2963805462 | Deep neural networks are becoming popular and important assets of many AI companies. However, recent studies indicate that they are also vulnerable to adversarial attacks. Adversarial attacks can be either white-box or black-box. The white-box attacks assume full knowledge of the models while the black-box ones assume ... | Naturally, this powerful attacking method can be applied to reveal DNN architectures or some related information. @cite_51 used timing side channels to infer the depth of the network; @cite_2 show that side channel attacks can roughly obtain information on activation functions, number of network layers, number of neuro... | {
"cite_N": [
"@cite_24",
"@cite_51",
"@cite_2"
],
"mid": [
"2789993878",
"2906869444",
"2964285340"
],
"abstract": [
"Deep learning has become the de-facto computational paradigm for various kinds of perception problems, including many privacy-sensitive applications such as online... |
1907.10343 | 2963426884 | Conventional object detection methods essentially suppose that the training and testing data are collected from a restricted target domain with expensive labeling cost. For alleviating the problem of domain dependency and cumbersome labeling, this paper proposes to detect objects in an unrestricted environment by lever... | The object detection is a basic task in computer vision and has been widely studied for many years. The earlier work @cite_34 @cite_10 @cite_33 of the object detection were implemented with sliding windows and boost classifiers. Benefited by the success of CNN models @cite_8 @cite_3 @cite_0 , a number of CNN based obje... | {
"cite_N": [
"@cite_31",
"@cite_33",
"@cite_8",
"@cite_28",
"@cite_41",
"@cite_29",
"@cite_3",
"@cite_16",
"@cite_0",
"@cite_24",
"@cite_2",
"@cite_15",
"@cite_34",
"@cite_10",
"@cite_20"
],
"mid": [
"2779991293",
"2125556102",
"2194775991",... |
1901.10760 | 2919401041 | This paper presents a novel clustering concept that is based on jointly learned nonlinear transforms (NTs) with priors on the information loss and the discrimination. We introduce a clustering principle that is based on evaluation of a parametric min-max measure for the discriminative prior. The decomposition of the pr... | Factor analysis @cite_7 and matrix factorization @cite_8 relay on decomposition on hidden features without or with constraints. One special case with only a constraint on the sparsity of the hidden representation, which is considered as a "hard" assignment is the basic k-means @cite_13 algorithm. When discrimination co... | {
"cite_N": [
"@cite_37",
"@cite_26",
"@cite_22",
"@cite_7",
"@cite_8",
"@cite_36",
"@cite_41",
"@cite_21",
"@cite_1",
"@cite_6",
"@cite_23",
"@cite_5",
"@cite_34",
"@cite_13",
"@cite_12"
],
"mid": [
"2055588122",
"",
"1986007546",
"19873... |
1901.10710 | 2950181282 | This paper proposes a novel training scheme for fast matching models in Search Ads, which is motivated by the real challenges in model training. The first challenge stems from the pursuit of high throughput, which prohibits the deployment of inseparable architectures, and hence greatly limits the model accuracy. The se... | As a subproblem of information retrieval, web search has long been an active research area, and has entailed a large body of literature. Methods in this area can be roughly grouped into two categories, namely traditional approaches and deep learning based models. Representative methods falling into the former category ... | {
"cite_N": [
"@cite_3",
"@cite_5",
"@cite_20",
"@cite_18",
"@cite_4",
"@cite_8",
"@cite_21",
"@cite_23",
"@cite_17",
"@cite_26",
"@cite_7",
"@cite_28",
"@cite_6",
"@cite_19",
"@cite_27",
"@cite_12",
"@cite_14",
"@cite_1",
"@cite_24",
"@c... |
1901.10710 | 2950181282 | This paper proposes a novel training scheme for fast matching models in Search Ads, which is motivated by the real challenges in model training. The first challenge stems from the pursuit of high throughput, which prohibits the deployment of inseparable architectures, and hence greatly limits the model accuracy. The se... | In particular, the authors of @cite_3 propose CDSSM by extending DSSM @cite_17 . Their CDSSM employs a convolutional layer as a sliding window based local feature extractor, and uses dimension-wise max-pooling to form the global semantic representation by combining all local features. Both DSSM and CDSSM are Siamese ne... | {
"cite_N": [
"@cite_9",
"@cite_3",
"@cite_17"
],
"mid": [
"2517540742",
"2186845332",
"2136189984"
],
"abstract": [
"Manually crafted combinatorial features have been the \"secret sauce\" behind many successful models. For web-scale applications, however, the variety and volume of... |
1901.10710 | 2950181282 | This paper proposes a novel training scheme for fast matching models in Search Ads, which is motivated by the real challenges in model training. The first challenge stems from the pursuit of high throughput, which prohibits the deployment of inseparable architectures, and hence greatly limits the model accuracy. The se... | Attention mechanisms have gained great popularity recently @cite_2 @cite_0 , which allows modeling of dependencies without regard to distance in the sequence. In particular, self-attention refers to an attention mechanism that relates different positions in a single sequence when extracting sequence representation, and... | {
"cite_N": [
"@cite_0",
"@cite_10",
"@cite_13",
"@cite_2"
],
"mid": [
"2413794162",
"2626778328",
"2597655663",
"2133564696"
],
"abstract": [
"",
"The dominant sequence transduction models are based on complex recurrent or convolutional neural networks in an encoder-de... |
1901.10710 | 2950181282 | This paper proposes a novel training scheme for fast matching models in Search Ads, which is motivated by the real challenges in model training. The first challenge stems from the pursuit of high throughput, which prohibits the deployment of inseparable architectures, and hence greatly limits the model accuracy. The se... | Finally, the idea of augmenting training data using unsupervised or weakly supervised auxiliary information has been adopted to help many tasks across different areas. One recent study implementing this idea is reported in @cite_15 , which investigates the potential of pretraining on hashtags of social media images ins... | {
"cite_N": [
"@cite_15"
],
"mid": [
"2963703197"
],
"abstract": [
"State-of-the-art visual perception models for a wide range of tasks rely on supervised pretraining. ImageNet classification is the de facto pretraining task for these models. Yet, ImageNet is now nearly ten years old and is by mod... |
1901.10401 | 2914836496 | This paper proposes a framework to analyze an emerging wireless architecture where vehicles collect data from devices. Using stochastic geometry, the devices are modeled by a planar Poisson point process. Independently, roads and vehicles are modeled by a Poisson line process and a Cox point process, respectively. For ... | The proposed network architecture is an example of random mobile ad hoc network or device-to-device network in the sense that it can expand the limited coverage of infrastructure or enable high-speed and low-distance communication between devices without infrastructure @cite_14 @cite_43 @cite_38 @cite_36 @cite_20 @cite... | {
"cite_N": [
"@cite_38",
"@cite_14",
"@cite_4",
"@cite_36",
"@cite_29",
"@cite_42",
"@cite_43",
"@cite_40",
"@cite_27",
"@cite_2",
"@cite_15",
"@cite_13",
"@cite_20"
],
"mid": [
"2140656373",
"2107846430",
"2095796369",
"1988834815",
"214587... |
1901.10401 | 2914836496 | This paper proposes a framework to analyze an emerging wireless architecture where vehicles collect data from devices. Using stochastic geometry, the devices are modeled by a planar Poisson point process. Independently, roads and vehicles are modeled by a Poisson line process and a Cox point process, respectively. For ... | However, modeling the locations of vehicles as a planar Poisson point process is inaccurate since almost surely no more than two points can be found on a line in the planar Poisson point process @cite_21 , and yet the locations of vehicles exhibit a linear pattern when they are on the same straight road. In order to ad... | {
"cite_N": [
"@cite_26",
"@cite_33",
"@cite_9",
"@cite_21",
"@cite_45"
],
"mid": [
"120871837",
"2963396489",
"2041078943",
"1541118220",
"2963692345"
],
"abstract": [
"",
"This paper analyzes an emerging architecture of cellular network utilizing both planar b... |
1901.10401 | 2914836496 | This paper proposes a framework to analyze an emerging wireless architecture where vehicles collect data from devices. Using stochastic geometry, the devices are modeled by a planar Poisson point process. Independently, roads and vehicles are modeled by a Poisson line process and a Cox point process, respectively. For ... | On the other hand, since vehicles are assumed to cover a wide area as they move on roads, it is essential to analyze the network behavior over time. This paper uses the theory of random closed sets @cite_35 @cite_3 to derive the area fractions of the coverage disks and of the progress of coverage over time, respectivel... | {
"cite_N": [
"@cite_35",
"@cite_31",
"@cite_22",
"@cite_28",
"@cite_1",
"@cite_3",
"@cite_39",
"@cite_6",
"@cite_46",
"@cite_10"
],
"mid": [
"",
"2149959815",
"2153422903",
"2130276881",
"2125957038",
"2118166339",
"2151824015",
"2149429274"... |
1901.10265 | 2912727640 | Case studies, such as , 2015 have shown that in image summarization, such as with Google Image Search, the people in the results presented for occupations are more imbalanced with respect to sensitive attributes such as gender and ethnicity than the ground truth. Most of the existing approaches to correct for this prob... | The study by @cite_22 explored the effects of bias in image search results of occupations on the perception of people of that occupation. The major aim of the study was to understand whether the biased portrayal of minorities in image search results leads to stereotypes or not. Such a phenomenon has been observed in ot... | {
"cite_N": [
"@cite_12",
"@cite_22",
"@cite_20",
"@cite_17"
],
"mid": [
"2250559305",
"2149252982",
"1964670863",
"2788481061"
],
"abstract": [
"Many NLP tools for English and German are based on manually annotated articles from the Wall Street Journal and Frankfurter Rund... |
1901.10265 | 2912727640 | Case studies, such as , 2015 have shown that in image summarization, such as with Google Image Search, the people in the results presented for occupations are more imbalanced with respect to sensitive attributes such as gender and ethnicity than the ground truth. Most of the existing approaches to correct for this prob... | There are other works that attempt to ensure diversity in the learning algorithm without using gender or race labels. @cite_23 consider the problem of gender bias in word embeddings trained on Google News articles and provide methods to modify the embeddings to debias them. Through standard gender-related words, they i... | {
"cite_N": [
"@cite_23"
],
"mid": [
"2483215953"
],
"abstract": [
"The blind application of machine learning runs the risk of amplifying biases present in data. Such a danger is facing us with word embedding, a popular framework to represent text data as vectors which has been used in many machin... |
1901.10265 | 2912727640 | Case studies, such as , 2015 have shown that in image summarization, such as with Google Image Search, the people in the results presented for occupations are more imbalanced with respect to sensitive attributes such as gender and ethnicity than the ground truth. Most of the existing approaches to correct for this prob... | To identify image similarity, a number of techniques have been explored @cite_25 . The usual techniques before the use of Convolutional Neural Networks included the following: Blob detection @cite_15 , involves finding the part of the image which is consistent across all images, Template matching @cite_7 , where we are... | {
"cite_N": [
"@cite_9",
"@cite_15",
"@cite_25",
"@cite_7"
],
"mid": [
"1677409904",
"2012231760",
"",
"2400360455"
],
"abstract": [
"In this paper, we present a novel scale- and rotation-invariant interest point detector and descriptor, coined SURF (Speeded Up Robust Featu... |
1901.10265 | 2912727640 | Case studies, such as , 2015 have shown that in image summarization, such as with Google Image Search, the people in the results presented for occupations are more imbalanced with respect to sensitive attributes such as gender and ethnicity than the ground truth. Most of the existing approaches to correct for this prob... | This method of using pre-trained models for other tasks is also called transfer learning''. This technique has been used in many other classification tasks, such thoraco-abdominal lymph node detection and interstitial lung disease classification @cite_2 , or object and action classification @cite_1 , and has shown sign... | {
"cite_N": [
"@cite_1",
"@cite_2"
],
"mid": [
"2161381512",
"2253429366"
],
"abstract": [
"Convolutional neural networks (CNN) have recently shown outstanding image classification performance in the large- scale visual recognition challenge (ILSVRC2012). The suc- cess of CNNs is attribute... |
1901.10240 | 2914485273 | Style transfer is a technique for combining two images based on the activations and feature statistics in a deep learning neural network architecture. This paper studies the analogous task in the audio domain and takes a critical look at the problems that arise when adapting the original vision-based framework to handl... | If articulating the intuitive distinction between style and content for images is difficult, it is even more so for sound, in particular non-speech sound. The recent use of neural network-derived statistics to describe such perceptual qualities for texture synthesis @cite_0 and style transfer @cite_22 offers a fresh co... | {
"cite_N": [
"@cite_0",
"@cite_22"
],
"mid": [
"2964193438",
"1924619199"
],
"abstract": [
"Here we introduce a new model of natural textures based on the feature spaces of convolutional neural networks optimised for object recognition. Samples from the model are of high perceptual qualit... |
1901.10240 | 2914485273 | Style transfer is a technique for combining two images based on the activations and feature statistics in a deep learning neural network architecture. This paper studies the analogous task in the audio domain and takes a critical look at the problems that arise when adapting the original vision-based framework to handl... | While all the related works studied certain aspects of the problem, none go into detail on the challenges posed by the nature of sound and how it is represented, especially in relation to what is essentially a vision inspired model in the CNN. Hence, the focus of this paper is not a presentation of state-of-the-art aud... | {
"cite_N": [
"@cite_22"
],
"mid": [
"1924619199"
],
"abstract": [
"In fine art, especially painting, humans have mastered the skill to create unique visual experiences through composing a complex interplay between the content and style of an image. Thus far the algorithmic basis of this process i... |
1901.10441 | 2912162169 | Obtaining sound inferences over remote networks via active or passive measurements is difficult. Active measurement campaigns face challenges of load, coverage, and visibility. Passive measurements require a privileged vantage point. Even networks under our own control too often remain poorly understood and hard to dia... | Significant prior literature focuses on performing passive measurement inferences. In addition to legitimate traffic, non-trivial levels of background radiation'' @cite_41 arrive at networks due to self-propagating malware, security scanners, and attacks. Casado al show the wealth of information that can be gleaned pas... | {
"cite_N": [
"@cite_38",
"@cite_37",
"@cite_7",
"@cite_41",
"@cite_9",
"@cite_12"
],
"mid": [
"2291500023",
"148447563",
"2095761072",
"2000497232",
"2465314304",
"2167994609"
],
"abstract": [
"The performance of Internet services is intrinsically tied to p... |
1901.10441 | 2912162169 | Obtaining sound inferences over remote networks via active or passive measurements is difficult. Active measurement campaigns face challenges of load, coverage, and visibility. Passive measurements require a privileged vantage point. Even networks under our own control too often remain poorly understood and hard to dia... | Individual networks frequently perform regular pair-wise measurements between nodes or networks under their control, Content Distribution Networks (CDNs) that run continual measurements to detect and route around path problems @cite_39 . The IHB seeks to push such functionality deeper into the network stack such that a... | {
"cite_N": [
"@cite_39"
],
"mid": [
"1976986662"
],
"abstract": [
"Comprising more than 61,000 servers located across nearly 1,000 networks in 70 countries worldwide, the Akamai platform delivers hundreds of billions of Internet interactions daily, helping thousands of enterprises boost the perfo... |
1901.10435 | 2919910909 | The article proposes a new framework for assessment of physical rehabilitation exercises based on a deep learning approach. The objective of the framework is automated quantification of patient performance in completing prescribed rehabilitation exercises, based on captured whole-body joint trajectories. The main compo... | Conventional approaches for mathematical modeling and representation of human movements are broadly classified into two categories: top-down approaches that introduce latent states for describing the temporal dynamics of the movements, and bottom-up approaches that employ local features for representing the movements. ... | {
"cite_N": [
"@cite_22",
"@cite_36",
"@cite_3",
"@cite_0",
"@cite_13"
],
"mid": [
"2319894388",
"2168162876",
"2016418141",
"2040378481",
"2054661770"
],
"abstract": [
"Mobility improvement for patients is one of the primary concerns of physiotherapy rehabilitation... |
1901.10435 | 2919910909 | The article proposes a new framework for assessment of physical rehabilitation exercises based on a deep learning approach. The objective of the framework is automated quantification of patient performance in completing prescribed rehabilitation exercises, based on captured whole-body joint trajectories. The main compo... | Recent developments in artificial NNs stirred significant interest in their application for modeling and analysis of human motions. Numerous works employed NNs for motion classification and applied the trained models for activity recognition, gait identification, gesture recognition, action localization, and related ap... | {
"cite_N": [
"@cite_4",
"@cite_33",
"@cite_15",
"@cite_41",
"@cite_9",
"@cite_6",
"@cite_39",
"@cite_24",
"@cite_19",
"@cite_27",
"@cite_31",
"@cite_16"
],
"mid": [
"28988658",
"2505121268",
"2950568498",
"2594167370",
"2176353499",
"2270470... |
1901.10435 | 2919910909 | The article proposes a new framework for assessment of physical rehabilitation exercises based on a deep learning approach. The objective of the framework is automated quantification of patient performance in completing prescribed rehabilitation exercises, based on captured whole-body joint trajectories. The main compo... | Several studies in the literature on exercise evaluation employed machine learning methods to classify the individual repetitions into of movements. Methods used for this purpose include Adaboost classifier @cite_18 , @math -nearest neighbors @cite_30 , Bayesian classifier @cite_26 , and an ensemble of multi-layer perc... | {
"cite_N": [
"@cite_30",
"@cite_18",
"@cite_26",
"@cite_32"
],
"mid": [
"",
"1990182801",
"2032984112",
"1977420570"
],
"abstract": [
"",
"In this paper, we describe methods for assessment of exercise quality using body-worn tri-axial accelerometers. We assess exercise... |
1901.10435 | 2919910909 | The article proposes a new framework for assessment of physical rehabilitation exercises based on a deep learning approach. The objective of the framework is automated quantification of patient performance in completing prescribed rehabilitation exercises, based on captured whole-body joint trajectories. The main compo... | The majority of related studies employed for deriving movement quality scores. Concretely, @cite_22 used a variant of the Mahalanobis distance to quantify the level of correctness of rehabilitation movements, based on a calculated distance between patient-performed repetitions and a set of repetitions performed by a gr... | {
"cite_N": [
"@cite_35",
"@cite_34",
"@cite_22",
"@cite_17"
],
"mid": [
"2128160875",
"1985745673",
"2319894388",
""
],
"abstract": [
"This paper reports on an optimum dynamic progxamming (DP) based time-normalization algorithm for spoken word recognition. First, a general... |
1901.10435 | 2919910909 | The article proposes a new framework for assessment of physical rehabilitation exercises based on a deep learning approach. The objective of the framework is automated quantification of patient performance in completing prescribed rehabilitation exercises, based on captured whole-body joint trajectories. The main compo... | Another body of research work utilized for modeling and evaluation of rehabilitation movements. Studies based on hidden Markov models @cite_14 , @cite_37 and mixtures of Gaussian distributions @cite_39 typically perform a quality assessment based on the likelihood that the individual sequences are being drawn from a tr... | {
"cite_N": [
"@cite_37",
"@cite_14",
"@cite_39"
],
"mid": [
"2343142751",
"2778538683",
"2580770992"
],
"abstract": [
"Movement primitive segmentation enables long sequences of human movement observation data to be segmented into smaller components, termed movement primitives , to... |
1901.10443 | 2911457688 | Motivated by concerns that machine learning algorithms may introduce significant bias in classification models, developing fair classifiers has become an important problem in machine learning research. One important paradigm towards this has been providing algorithms for adversarially learning fair classifiers (, 2018;... | The idea of adversarial machine learning was popularized by the introduction of Generative Adversarial Networks (GANs) @cite_13 . Based on similar ideas, multiple learning algorithms have been suggested to generate fair classifiers using adversaries. | {
"cite_N": [
"@cite_13"
],
"mid": [
"2099471712"
],
"abstract": [
"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 estim... |
1901.10443 | 2911457688 | Motivated by concerns that machine learning algorithms may introduce significant bias in classification models, developing fair classifiers has become an important problem in machine learning research. One important paradigm towards this has been providing algorithms for adversarially learning fair classifiers (, 2018;... | As mentioned earlier, @cite_0 proposed a model to learn a fair classifier based on the idea of adversarial debiasing. Their algorithm also uses the gradient descent with the modified update but does not include any theoretical guarantees on convergence, and experimentally, they suggest a model to just ensure equalized ... | {
"cite_N": [
"@cite_0",
"@cite_27",
"@cite_34"
],
"mid": [
"2784397426",
"",
"2810290439"
],
"abstract": [
"Machine learning is a tool for building models that accurately represent input training data. When undesired biases concerning demographic groups are in the training data, w... |
1901.10443 | 2911457688 | Motivated by concerns that machine learning algorithms may introduce significant bias in classification models, developing fair classifiers has become an important problem in machine learning research. One important paradigm towards this has been providing algorithms for adversarially learning fair classifiers (, 2018;... | The work of @cite_31 is perhaps the closest in terms of the techniques involved. They formulate their constrained optimization problem as an unconstrained one using Lagrangian transformation. This leads to a min-max optimization problem, which they then solve using the saddle point methods of @cite_9 @cite_17 . The key... | {
"cite_N": [
"@cite_9",
"@cite_31",
"@cite_17"
],
"mid": [
"2097477220",
"2790744245",
""
],
"abstract": [
"We study the close connections between game theory, on-line prediction and boosting. After a brief review of game theory, we describe an algorithm for learning to play repea... |
1901.10423 | 2911776884 | We present a decentralized algorithm to achieve segregation into an arbitrary number of groups with swarms of autonomous robots. The distinguishing feature of our approach is in the minimalistic assumptions on which it is based. Specifically, we assume that (i) Each robot is equipped with a ternary sensor capable of de... | Segregation is a common behavior in nature, and it can be observed across scales. For example, cell segregation is a basic building block of embryogeneis in tissue generation processes @cite_3 @cite_0 ; while social insects, such as ants, organize their brood into ring-like structures @cite_14 . | {
"cite_N": [
"@cite_0",
"@cite_14",
"@cite_3"
],
"mid": [
"1977973890",
"2001868359",
"2103852413"
],
"abstract": [
"",
"Leptothorax unifasciatus ant colonies occupy flat crevices in rocks in which their brood is kept in a single cluster. In artificial nests made from two glas... |
1901.10423 | 2911776884 | We present a decentralized algorithm to achieve segregation into an arbitrary number of groups with swarms of autonomous robots. The distinguishing feature of our approach is in the minimalistic assumptions on which it is based. Specifically, we assume that (i) Each robot is equipped with a ternary sensor capable of de... | In robotics, segregation is a problem that has not received considerable attention. The main methods that have been proposed so far are based on some variation of the artificial potential approach @cite_17 , which assumes that the robots can detect each other and estimate relative distance vectors. | {
"cite_N": [
"@cite_17"
],
"mid": [
"1976486181"
],
"abstract": [
"We introduce a framework, called “physicomimetics,” that provides distributed control of large collections of mobile physical agents in sensor networks. The agents sense and react to virtual forces, which are motivated by natural ... |
1901.10423 | 2911776884 | We present a decentralized algorithm to achieve segregation into an arbitrary number of groups with swarms of autonomous robots. The distinguishing feature of our approach is in the minimalistic assumptions on which it is based. Specifically, we assume that (i) Each robot is equipped with a ternary sensor capable of de... | Gro @cite_12 proposed an algorithm inspired by the Brazil Nut effect, in which the robots form regular layers simulating gravity by sharing a common direction. This study was later extended to work on e-pucks robots @cite_18 . To simulate gravity, this approach requires the robots to share a common target vector, which... | {
"cite_N": [
"@cite_18",
"@cite_12"
],
"mid": [
"1686337294",
"2160636171"
],
"abstract": [
"When a mixture of particles with different attributes undergoes vibration, a segregation pattern is often observed. For example, in muesli cereal packs, the largest particles---the Brazil nuts---t... |
1901.10423 | 2911776884 | We present a decentralized algorithm to achieve segregation into an arbitrary number of groups with swarms of autonomous robots. The distinguishing feature of our approach is in the minimalistic assumptions on which it is based. Specifically, we assume that (i) Each robot is equipped with a ternary sensor capable of de... | Kumar @cite_6 introduced the concept of differential potential'', whereby two robots experience a different artificial potential depending on their being part of the same class or not. The convergence of this approach is guaranteed for two classes, but when more classes are employed local minima prevent segregation fro... | {
"cite_N": [
"@cite_6"
],
"mid": [
"2123457819"
],
"abstract": [
"There are several examples in natural systems that exhibit the self-organizing behavior of segregation when different types of units interact with each other. One of the best examples is a system of biological cells of heterogeneou... |
1901.10469 | 2770825526 | Near Earth Asteroids (NEAs) are discovered daily, mainly by few major surveys, nevertheless many of them remain unobserved for years, even decades. Even so, there is room for new discoveries, including those submitted by smaller projects and amateur astronomers. Besides the well-known surveys that have their own automa... | Variable KD-Tree algorithms were used by the Pan-STARRS Moving Object Processing System @cite_0 , @cite_7 . These techniques were developed in collaboration with the incoming LSST (Large Synoptic Survey Telescope) @cite_3 . | {
"cite_N": [
"@cite_0",
"@cite_3",
"@cite_7"
],
"mid": [
"2164633392",
"84659944",
"2123353273"
],
"abstract": [
"In this paper we consider the problem of finding sets of points that conform to a given underlying model from within a dense, noisy set of observations. This problem i... |
1901.10469 | 2770825526 | Near Earth Asteroids (NEAs) are discovered daily, mainly by few major surveys, nevertheless many of them remain unobserved for years, even decades. Even so, there is room for new discoveries, including those submitted by smaller projects and amateur astronomers. Besides the well-known surveys that have their own automa... | Not only the major surveys developed automated system for moving object detection, but also some amateurs and small private surveys. A group of mainly amateurs search for NEAs in the TOTAS survey carried out with the ESA-OGS 1m telescope, lead by ESA @cite_4 . Another group from Argentina developed such a system based ... | {
"cite_N": [
"@cite_1",
"@cite_4"
],
"mid": [
"197143460",
"2251089998"
],
"abstract": [
"In this work we present a system for autonomous discovery of asteroids, space trash and other moving objects. This system performs astronomical image data reduction based on an image processing pipel... |
1901.09960 | 2951458896 | Tuning a pre-trained network is commonly thought to improve data efficiency. However, Kaiming have called into question the utility of pre-training by showing that training from scratch can often yield similar performance, should the model train long enough. We show that although pre-training may not improve performanc... | It is well-known that pre-training improves generalization when the dataset for the target task is extremely small. Prior work on transfer learning has analyzed the properties of this effect, such as when fine-tuning should stop @cite_4 and which layers should be fine-tuned @cite_27 . In a series of ablation studies, s... | {
"cite_N": [
"@cite_27",
"@cite_4"
],
"mid": [
"2149933564",
"2160921898"
],
"abstract": [
"Many deep neural networks trained on natural images exhibit a curious phenomenon in common: on the first layer they learn features similar to Gabor filters and color blobs. Such first-layer feature... |
1901.09960 | 2951458896 | Tuning a pre-trained network is commonly thought to improve data efficiency. However, Kaiming have called into question the utility of pre-training by showing that training from scratch can often yield similar performance, should the model train long enough. We show that although pre-training may not improve performanc... | show that networks overfit to the incorrect labels when trained for too long ( fig:trainforlonger ). This observation suggests pre-training as a potential fix, since one need only fine-tune for a short period to attain good performance. We show that pre-training not only improves performance with no label noise correct... | {
"cite_N": [
"@cite_14",
"@cite_10",
"@cite_32",
"@cite_3"
],
"mid": [
"2964309657",
"2963735582",
"2964292098",
"2804077623"
],
"abstract": [
"The growing importance of massive datasets with the advent of deep learning makes robustness to label noise a critical property f... |
1901.09960 | 2951458896 | Tuning a pre-trained network is commonly thought to improve data efficiency. However, Kaiming have called into question the utility of pre-training by showing that training from scratch can often yield similar performance, should the model train long enough. We show that although pre-training may not improve performanc... | The susceptibility of neural networks to small, adversarially chosen input perturbations has received much attention. Over the years, many methods have been proposed as defenses against adversarial examples @cite_24 @cite_12 , but these are often circumvented in short order @cite_18 . In fact, the only defense widely r... | {
"cite_N": [
"@cite_24",
"@cite_18",
"@cite_21",
"@cite_12"
],
"mid": [
"2593892853",
"2619479788",
"2510153535",
"2963695663"
],
"abstract": [
"Machine learning and deep learning in particular has advanced tremendously on perceptual tasks in recent years. However, it rema... |
1901.10159 | 2950990113 | To understand the dynamics of optimization in deep neural networks, we develop a tool to study the evolution of the entire Hessian spectrum throughout the optimization process. Using this, we study a number of hypotheses concerning smoothness, curvature, and sharpness in the deep learning literature. We then thoroughly... | During the preparation of this paper, @cite_29 appeared on Arxiv which briefly introduces the same spectrum estimation methodology and studies the Hessian on small subsamples of MNIST and CIFAR-10 at the end of the training. In comparison, we provide a detailed exposition, error analysis and validation of the estimator... | {
"cite_N": [
"@cite_29"
],
"mid": [
"2900531695"
],
"abstract": [
"Previous works observed the spectrum of the Hessian of the training loss of deep neural networks. However, the networks considered were of minuscule size. We apply state-of-the-art tools in modern high-dimensional numerical linear... |
1901.10172 | 2913075819 | In this paper, we present a two-stream multi-task network for fashion recognition. This task is challenging as fashion clothing always contain multiple attributes, which need to be predicted simultaneously for real-time industrial systems. To handle these challenges, we formulate fashion recognition into a multi-task l... | has been widely studied in recent years. As a general and important computer vision task, it composes far-reaching applications such as clothing retrieval @cite_3 , recognition @cite_7 , fashion landmark detection @cite_12 , and clothing recommendation @cite_4 . To solve this task, early methods @cite_2 heavily rely on... | {
"cite_N": [
"@cite_4",
"@cite_7",
"@cite_3",
"@cite_2",
"@cite_12"
],
"mid": [
"2737102415",
"2471768434",
"2200092826",
"146395692",
"2511502099"
],
"abstract": [
"The ubiquity of online fashion shopping demands effective recommendation services for customers. In... |
1901.10172 | 2913075819 | In this paper, we present a two-stream multi-task network for fashion recognition. This task is challenging as fashion clothing always contain multiple attributes, which need to be predicted simultaneously for real-time industrial systems. To handle these challenges, we formulate fashion recognition into a multi-task l... | In @cite_18 , the authors leveraged a dual attribute aware mechanism for clothing retrieval. Differently, Liu @cite_7 presented a multi-branch network for clothing classification, retrieval, and landmark detection. @cite_5 utilized a model to precisely localize attribute for fashion search. More recently, Wang @cite_14... | {
"cite_N": [
"@cite_5",
"@cite_18",
"@cite_14",
"@cite_7"
],
"mid": [
"2798951647",
"2170881581",
"2798734012",
"2471768434"
],
"abstract": [
"In this paper, we investigate ways of conducting a detailed fashion search using query images and attributes. A credible fashion s... |
1901.10172 | 2913075819 | In this paper, we present a two-stream multi-task network for fashion recognition. This task is challenging as fashion clothing always contain multiple attributes, which need to be predicted simultaneously for real-time industrial systems. To handle these challenges, we formulate fashion recognition into a multi-task l... | has shown promising results in many applications. A comprehensive survey can be found in @cite_17 . In consideration of brevity, we hereby only introduce related MTL literature focusing on computer vision tasks. @cite_0 introduced a multi-task deep convolution neural network to jointly achieve body-part and joint-point... | {
"cite_N": [
"@cite_9",
"@cite_1",
"@cite_6",
"@cite_0",
"@cite_17"
],
"mid": [
"1907729166",
"2213713716",
"2199851682",
"2078224158",
"2742079690"
],
"abstract": [
"This paper proposes a joint multi-task learning algorithm to better predict attributes in images u... |
1901.10139 | 2913967867 | It is an easy task for humans to learn and generalize a problem, perhaps it is due to their ability to visualize and imagine unseen objects and concepts. The power of imagination comes handy especially when interpolating learnt experience (like seen examples) over new classes of a problem. For a machine learning system... | The goal of few shot learning is to learn a representation which can generalize across classes and deal with even unseen examples from new classes @cite_4 . The few shot problem specifically has been studied from multiple perspectives, including the optimization @cite_18 , metric learning @cite_19 , similarity-matching... | {
"cite_N": [
"@cite_18",
"@cite_26",
"@cite_4",
"@cite_21",
"@cite_1",
"@cite_17",
"@cite_19",
"@cite_13",
"@cite_11"
],
"mid": [
"2753160622",
"2562986790",
"2194321275",
"2783820753",
"2133774033",
"",
"2963341924",
"2963636184",
"21503854... |
1901.10139 | 2913967867 | It is an easy task for humans to learn and generalize a problem, perhaps it is due to their ability to visualize and imagine unseen objects and concepts. The power of imagination comes handy especially when interpolating learnt experience (like seen examples) over new classes of a problem. For a machine learning system... | The goal of few shot learning is to learn a representation which can generalize across classes and deal with even unseen examples from new classes @cite_4 . The few shot problem specifically has been studied from multiple perspectives, including optimization @cite_18 , metric learning @cite_19 , similarity-matching @ci... | {
"cite_N": [
"@cite_18",
"@cite_26",
"@cite_4",
"@cite_21",
"@cite_1",
"@cite_17",
"@cite_19",
"@cite_13",
"@cite_11"
],
"mid": [
"2753160622",
"2562986790",
"2194321275",
"2783820753",
"2133774033",
"",
"2963341924",
"2963636184",
"21503854... |
1901.10173 | 2912152488 | Recent studies have shown that imbalance ratio is not the only cause of the performance loss of a classifier in imbalanced data classification. In fact, other data factors, such as small disjuncts, noises and overlapping, also play the roles in tandem with imbalance ratio, which makes the problem difficult. Thus far, t... | The second data factor is noise. Noisy samples are usually defined as the ones from one class located deep into the other class @cite_17 . The existence of noise samples in the minority class will make blind oversampling methods like SMOTE generate more noises, so that applying oversampling on the noisy the minority cl... | {
"cite_N": [
"@cite_9",
"@cite_0",
"@cite_2",
"@cite_34",
"@cite_13",
"@cite_17"
],
"mid": [
"1993220166",
"1592804209",
"",
"2119168155",
"2137029138",
"85350352"
],
"abstract": [
"There are several aspects that might influence the performance achieved by ... |
1901.10173 | 2912152488 | Recent studies have shown that imbalance ratio is not the only cause of the performance loss of a classifier in imbalanced data classification. In fact, other data factors, such as small disjuncts, noises and overlapping, also play the roles in tandem with imbalance ratio, which makes the problem difficult. Thus far, t... | Before we close this section, we would like to point out that another somewhat related area is data complexity. A list of complexity measures are proposed in @cite_28 with different featured groups. The measures are used to study the essential structure of data and guide classifier selection for specific problems. Rece... | {
"cite_N": [
"@cite_28",
"@cite_16"
],
"mid": [
"2125877832",
"2022477494"
],
"abstract": [
"We studied a number of measures that characterize the difficulty of a classification problem, focusing on the geometrical complexity of the class boundary. We compared a set of real-world problems... |
1901.09955 | 2913996813 | The crossing number of a graph @math is the least number of crossings over all possible drawings of @math . We present a structural characterization of graphs with crossing number one. | The problem of characterizing graphs with crossing number at least two was already studied by Arroyo and Richter @cite_0 in the context of . | {
"cite_N": [
"@cite_0"
],
"mid": [
"2557594551"
],
"abstract": [
"Our main result includes the following, slightly surprising, fact: a 4-connected nonplanar graph G has crossing number at least 2 if and only if, for every pair e,f of edges having no common incident vertex, there are vertex-disjoi... |
1901.09955 | 2913996813 | The crossing number of a graph @math is the least number of crossings over all possible drawings of @math . We present a structural characterization of graphs with crossing number one. | Akka, Jendrol, Kle s c , and Panshetty @cite_15 obtained a characterization of planar graphs whose line graph has crossing number two. | {
"cite_N": [
"@cite_15"
],
"mid": [
"2049597710"
],
"abstract": [
"In this paper we deduce a necessary and sufficient condition for a line grah to have crossing number 1. In addition, we prove that the line graph of any nonplanar graph has crossing number greater than 2."
]
} |
1901.09955 | 2913996813 | The crossing number of a graph @math is the least number of crossings over all possible drawings of @math . We present a structural characterization of graphs with crossing number one. | A great deal of attention has been given to 2-crossing-critical graphs @cite_3 @cite_11 @cite_13 @cite_8 @cite_6 @cite_4 @cite_12 . For a positive integer @math , the @math on @math vertices, is the graph obtained from a @math -cycle by joining vertices with distance @math in the cycle. Bokal, Opporowski, Richter and S... | {
"cite_N": [
"@cite_4",
"@cite_8",
"@cite_6",
"@cite_3",
"@cite_13",
"@cite_12",
"@cite_11"
],
"mid": [
"2112497773",
"2090034967",
"",
"193336535",
"2056837081",
"2507315201",
"2126687412"
],
"abstract": [
"A graph is crossing-critical if deleting ... |
1901.10109 | 2913416354 | Massive volumes of data continuously generated on social platforms have become an important information source for users. A primary method to obtain fresh and valuable information from social streams is . Although there have been extensive studies on social search, existing methods only focus on the of query results bu... | Keyword-based approaches @cite_39 @cite_24 @cite_15 @cite_23 @cite_28 @cite_35 @cite_10 @cite_33 typically define to retrieve @math elements with the highest scores as the results where the scoring functions combine the to query keywords (measured by TF-IDF or BM25) with other contexts such as @cite_15 @cite_28 @cite_3... | {
"cite_N": [
"@cite_35",
"@cite_33",
"@cite_28",
"@cite_39",
"@cite_24",
"@cite_23",
"@cite_15",
"@cite_10"
],
"mid": [
"2582814973",
"2616258542",
"1537450505",
"2076538837",
"2071080574",
"2055483934",
"",
"1994962537"
],
"abstract": [
"Th... |
1901.10109 | 2913416354 | Massive volumes of data continuously generated on social platforms have become an important information source for users. A primary method to obtain fresh and valuable information from social streams is . Although there have been extensive studies on social search, existing methods only focus on the of query results bu... | As the metrics for textual relevance cannot fully represent the semantic relevance between user interest and text, recent work @cite_18 @cite_17 introduces topic models @cite_41 into social search, where user queries and elements are modeled as vectors in the topic space. The relevance between a query and an element is... | {
"cite_N": [
"@cite_41",
"@cite_18",
"@cite_17"
],
"mid": [
"",
"2433793808",
"2615162663"
],
"abstract": [
"",
"Social media advertising is a multi-billion dollar market and has become the major revenue source for Facebook and Twitter. To deliver ads to potentially interested... |
1901.10109 | 2913416354 | Massive volumes of data continuously generated on social platforms have become an important information source for users. A primary method to obtain fresh and valuable information from social streams is . Although there have been extensive studies on social search, existing methods only focus on the of query results bu... | There have been extensive studies on social stream summarization @cite_30 @cite_26 @cite_22 @cite_19 @cite_6 @cite_25 @cite_31 @cite_32 : the problem of extracting a set of elements from social streams. @cite_26 @cite_22 propose a framework for social stream summarization based on dynamic clustering. @cite_32 focus on ... | {
"cite_N": [
"@cite_30",
"@cite_26",
"@cite_22",
"@cite_29",
"@cite_32",
"@cite_6",
"@cite_19",
"@cite_31",
"@cite_25"
],
"mid": [
"2605324157",
"2028544407",
"1997164082",
"2613626598",
"2102230199",
"2093541376",
"2104117327",
"2599637166",
... |
1901.10109 | 2913416354 | Massive volumes of data continuously generated on social platforms have become an important information source for users. A primary method to obtain fresh and valuable information from social streams is . Although there have been extensive studies on social search, existing methods only focus on the of query results bu... | Submodular maximization has attracted a lot of research interest recently for its theoretical significance and wide applications. The standard approaches to submodular maximization with a cardinality constraint are the greedy heuristic @cite_0 and its improved version CELF @cite_36 , both of which are @math -approximat... | {
"cite_N": [
"@cite_36",
"@cite_0",
"@cite_27",
"@cite_40",
"@cite_34",
"@cite_13",
"@cite_12"
],
"mid": [
"2141403143",
"2757107770",
"2252172643",
"",
"2101246692",
"2842078607",
"1997959284"
],
"abstract": [
"Given a water distribution network, w... |
1901.10076 | 2914865839 | We study the learnability of a class of compact operators known as Schatten--von Neumann operators. These operators between infinite-dimensional function spaces play a central role in a variety of applications in learning theory and inverse problems. We address the question of sample complexity of learning Schatten-von... | On the algorithmic side Abernethy @cite_10 propose learning algorithms for a problem related to ours. They show how in the context of collaborative filtering, a number of existing algorithms can be abstractly modeled as learning compact operators, and derive a representer theorem which casts the problem as optimization... | {
"cite_N": [
"@cite_10"
],
"mid": [
"2116413942"
],
"abstract": [
"We present a general approach for collaborative filtering (CF) using spectral regularization to learn linear operators mapping a set of \"users\" to a set of possibly desired \"objects\". In particular, several recent low-rank typ... |
1901.09953 | 2911736397 | Deep generative models have been successfully applied to many applications. However, existing works experience limitations when generating large images (the literature usually generates small images, e.g. 32 * 32 or 128 * 128). In this paper, we propose a novel scheme, called deep tensor adversarial generative nets (TG... | In order to generate high-resolution images from low-resolution images, the model SRGAN @cite_19 is proposed to realize super-resolution of images. It uses CNN for extracting features from low-resolution images. The model of SRGAN testifies the strong capability of generative models in applications of images super-reso... | {
"cite_N": [
"@cite_19",
"@cite_4",
"@cite_20"
],
"mid": [
"",
"2964024144",
"2173520492"
],
"abstract": [
"",
"Synthesizing high-quality images from text descriptions is a challenging problem in computer vision and has many practical applications. Samples generated by existin... |
1901.09953 | 2911736397 | Deep generative models have been successfully applied to many applications. However, existing works experience limitations when generating large images (the literature usually generates small images, e.g. 32 * 32 or 128 * 128). In this paper, we propose a novel scheme, called deep tensor adversarial generative nets (TG... | However, image representation in pixel space may not be an efficient way as in the traditional GANs. Tensor representation based methods have been adopted recently. Recent papers @cite_18 @cite_1 apply tensor representation for dictionary learning with smaller dictionary size and better results than the traditional met... | {
"cite_N": [
"@cite_18",
"@cite_1",
"@cite_7"
],
"mid": [
"2030351687",
"2788913644",
"2024165284"
],
"abstract": [
"In dynamic computed tomography (CT) reconstruction, the data acquisition speed limits the spatio-temporal resolution. Recently, compressed sensing theory has been i... |
1901.10080 | 2912501354 | We tackle the problem of algorithmic fairness, where the goal is to avoid the unfairly influence of sensitive information, in the general context of regression with possible continuous sensitive attributes. We extend the framework of fair empirical risk minimization to this general scenario, covering in this way the wh... | In the context of fairness, most of the papers in literature address the problem of binary classification task with categorical (or even binary) sensitive features @cite_17 @cite_38 ; a broad review on classification with categorical sensitive feature is provided in @cite_8 . This task is indeed very important, because... | {
"cite_N": [
"@cite_38",
"@cite_8",
"@cite_17"
],
"mid": [
"2540757487",
"2963803533",
"2530395818"
],
"abstract": [
"Automated data-driven decision making systems are increasingly being used to assist, or even replace humans in many settings. These systems function by learning fr... |
1901.10080 | 2912501354 | We tackle the problem of algorithmic fairness, where the goal is to avoid the unfairly influence of sensitive information, in the general context of regression with possible continuous sensitive attributes. We extend the framework of fair empirical risk minimization to this general scenario, covering in this way the wh... | Focusing on the works able to handle regression tasks, we can divide them by the type of problems they are able to solve and the notion of fairness they exploit. As we will see, with very few exceptions -- e.g. @cite_24 -- most of the methods in literature are not able to deal with both classification and regression ta... | {
"cite_N": [
"@cite_24"
],
"mid": [
"2766939712"
],
"abstract": [
"Algorithmic decision making process now affects many aspects of our lives. Standard tools for machine learning, such as classification and regression, are subject to the bias in data, and thus direct application of such off-the-sh... |
1901.10080 | 2912501354 | We tackle the problem of algorithmic fairness, where the goal is to avoid the unfairly influence of sensitive information, in the general context of regression with possible continuous sensitive attributes. We extend the framework of fair empirical risk minimization to this general scenario, covering in this way the wh... | The largest family of methods tackle regression problems with (single) categorical or binary sensitive feature @cite_29 @cite_5 @cite_30 @cite_14 . For example, in @cite_29 , a convex approach for regression is proposed, where the authors use a specific definition of fairness in order to have models which treat similar... | {
"cite_N": [
"@cite_30",
"@cite_14",
"@cite_29",
"@cite_1",
"@cite_5"
],
"mid": [
"2914415610",
"2779140735",
"2622808887",
"2100960835",
"2155982052"
],
"abstract": [
"Fairness, through its many forms and definitions, has become an important issue facing the machi... |
1901.10080 | 2912501354 | We tackle the problem of algorithmic fairness, where the goal is to avoid the unfairly influence of sensitive information, in the general context of regression with possible continuous sensitive attributes. We extend the framework of fair empirical risk minimization to this general scenario, covering in this way the wh... | Reducing the regression problem to have only categorical sensitive features is a serious limitation. In this sense, few interesting papers present regression methods able to deal with continuous sensitive attributes @cite_24 @cite_2 @cite_10 . Differently to our approach, the authors impose other definitions of fairnes... | {
"cite_N": [
"@cite_24",
"@cite_38",
"@cite_10",
"@cite_2"
],
"mid": [
"2766939712",
"2540757487",
"2963092241",
"2885501813"
],
"abstract": [
"Algorithmic decision making process now affects many aspects of our lives. Standard tools for machine learning, such as classific... |
1901.10073 | 2913888055 | Recovering class inheritance from C++ binaries has several security benefits including problems such as decompilation and program hardening. Thanks to the optimization guidelines prescribed by the C++ standard, commercial C++ binaries tend to be optimized. While state-of-the-art class inheritance inference solutions ar... | OBJDigger, presented by @cite_25 , uses symbolic execution and inter-procedural data flow analysis to recover object instances, data members and methods of the same class. This is achieved by tracking the usage and propagation of the within and between functions. While the authors did not attempt to recover class inher... | {
"cite_N": [
"@cite_25"
],
"mid": [
"2144981449"
],
"abstract": [
"Object-oriented programming complicates the already difficult task of reverse engineering software, and is being used increasingly by malware authors. Unlike traditional procedural-style code, reverse engineers must understand the... |
1901.10073 | 2913888055 | Recovering class inheritance from C++ binaries has several security benefits including problems such as decompilation and program hardening. Thanks to the optimization guidelines prescribed by the C++ standard, commercial C++ binaries tend to be optimized. While state-of-the-art class inheritance inference solutions ar... | @cite_13 presented SmartDec which partially recovers certain C++ specific language constructs statically. It attempts to recover classes and their inheritance, virtual and non-virtual member functions, calls to virtual functions, exception raising and handling statements. Its main limitation is the inability to differe... | {
"cite_N": [
"@cite_13"
],
"mid": [
"2154522949"
],
"abstract": [
"This paper presents a method for automatic reconstruction of polymorphic class hierarchies from the assembly code obtained by compiling a C++ program. If the program is compiled with run-time type information (RTTI), class hierarc... |
1901.10073 | 2913888055 | Recovering class inheritance from C++ binaries has several security benefits including problems such as decompilation and program hardening. Thanks to the optimization guidelines prescribed by the C++ standard, commercial C++ binaries tend to be optimized. While state-of-the-art class inheritance inference solutions ar... | Rewards @cite_19 is one of many (e.g., TIE @cite_27 , Laika @cite_18 ) data structure reverse engineering tools to infer type information from binaries. It uses dynamic analysis to recover syntax and semantics of data structures observed during execution. Rewards only attempts to infer primitive data types of variables... | {
"cite_N": [
"@cite_19",
"@cite_27",
"@cite_18"
],
"mid": [
"2504609973",
"191489030",
"182734301"
],
"abstract": [
"With only the binary executable of a program, it is useful to discover the program's data structures and infer their syntactic and semantic definitions. Such knowle... |
1901.10073 | 2913888055 | Recovering class inheritance from C++ binaries has several security benefits including problems such as decompilation and program hardening. Thanks to the optimization guidelines prescribed by the C++ standard, commercial C++ binaries tend to be optimized. While state-of-the-art class inheritance inference solutions ar... | OOAnalyzer @cite_5 mainly groups methods into classes by combining traditional binary analysis, symbolic analysis and Prolog-based reasoning. The paper explained that class size and VTable size can be considered to decide inheritance. Since OOAnalyzer also considers non-polymorphic classes, one would assume that class ... | {
"cite_N": [
"@cite_5"
],
"mid": [
"2890042297"
],
"abstract": [
"High-level C++ source code abstractions such as classes and methods greatly assist human analysts and automated algorithms alike when analyzing C++ programs. Unfortunately, these abstractions are lost when compiling C++ source code... |
1901.09858 | 2913816046 | Differential privacy mechanisms that also make reconstruction of the data impossible come at a cost - a decrease in utility. In this paper, we tackle this problem by designing a private data release mechanism that makes reconstruction of the original data impossible and also preserves utility for a wide range of machin... | @cite_22 developed a randomization mechanism that utilized the JL transform and the Gaussian mechanism @cite_28 to provide non-interactive differential privacy with respect to attribute changes. They showed that their mechanism preserved utility by preserving distances in expectation. However, a shortcoming of this app... | {
"cite_N": [
"@cite_28",
"@cite_22",
"@cite_23"
],
"mid": [
"2027595342",
"1602085912",
""
],
"abstract": [
"The problem of privacy-preserving data analysis has a long history spanning multiple disciplines. As electronic data about individuals becomes increasingly detailed, and as... |
1901.09858 | 2913816046 | Differential privacy mechanisms that also make reconstruction of the data impossible come at a cost - a decrease in utility. In this paper, we tackle this problem by designing a private data release mechanism that makes reconstruction of the original data impossible and also preserves utility for a wide range of machin... | @cite_17 showed that the JL transform itself preserved differential privacy and provided utility guarantees in the strict case when only the covariance matrix is released. However, in order to provide privacy guarantees, the data matrix was required to be full rank with eigenvalues above some threshold. Since this is n... | {
"cite_N": [
"@cite_26",
"@cite_17"
],
"mid": [
"2951231886",
"2949485285"
],
"abstract": [
"This work studies formal utility and privacy guarantees for a simple multiplicative database transformation, where the data are compressed by a random linear or affine transformation, reducing the... |
1901.09735 | 2914576932 | Rapid progress in genomics has enabled a thriving market for direct-to-consumer' genetic testing, whereby people have access to their genetic information without the involvement of a healthcare provider. Companies like 23andMe and AncestryDNA, which provide affordable health, genealogy, and ancestry reports, have alrea... | @cite_3 conduct a meta-analysis of 53 studies involving 47K people around perceptions of genetic privacy, highlighting how survey questions are often phrased poorly, thus leading to possible misinterpretations of the results. They also show that not enough attention was paid to influential factors, e.g., participants' ... | {
"cite_N": [
"@cite_30",
"@cite_35",
"@cite_33",
"@cite_65",
"@cite_3",
"@cite_46"
],
"mid": [
"2619042849",
"2018622279",
"2319348939",
"2149278098",
"2898683213",
"1588965447"
],
"abstract": [
"Background Genetic testing for risk of hereditary cancer can ... |
1901.09697 | 2966439069 | We propose Bayesian differential privacy, a relaxation of differential privacy for similarly distributed data, that provides sharper privacy guarantees in difficult scenarios, such as deep learning. We also derive a general privacy accounting method for iterative learning algorithms under Bayesian differential privacy ... | As machine learning applications become more and more common, various vulnerabilities and attacks on ML models get discovered (for example, model inversion @cite_35 and membership inference @cite_31 ), raising the need of developing matching defences. | {
"cite_N": [
"@cite_35",
"@cite_31"
],
"mid": [
"2051267297",
"2535690855"
],
"abstract": [
"Machine-learning (ML) algorithms are increasingly utilized in privacy-sensitive applications such as predicting lifestyle choices, making medical diagnoses, and facial recognition. In a model inve... |
1901.09697 | 2966439069 | We propose Bayesian differential privacy, a relaxation of differential privacy for similarly distributed data, that provides sharper privacy guarantees in difficult scenarios, such as deep learning. We also derive a general privacy accounting method for iterative learning algorithms under Bayesian differential privacy ... | Differential privacy @cite_16 @cite_36 is one of the strongest privacy standards that can be employed to protect ML models from these and other attacks. Since pure DP is hard to achieve in many realistic complex learning tasks, a notion of approximate @math -DP is used across-the-board in machine learning. It is often ... | {
"cite_N": [
"@cite_33",
"@cite_36",
"@cite_29",
"@cite_24",
"@cite_23",
"@cite_16",
"@cite_20"
],
"mid": [
"2130964760",
"2517104773",
"2809008139",
"",
"2963699739",
"2564029303",
"2294904676"
],
"abstract": [
"We examine the information-theoretic... |
1901.09697 | 2966439069 | We propose Bayesian differential privacy, a relaxation of differential privacy for similarly distributed data, that provides sharper privacy guarantees in difficult scenarios, such as deep learning. We also derive a general privacy accounting method for iterative learning algorithms under Bayesian differential privacy ... | Privacy analysis in the context of differentially private ML is often done and lies in finding parameters @math (or bounds on it) that apply to the entire learning process, as opposed to fixing @math beforehand and calibrating noise to satisfy it. Due to the nature of such analysis (keeping track and accumulating some ... | {
"cite_N": [
"@cite_33"
],
"mid": [
"2130964760"
],
"abstract": [
"We examine the information-theoretic foundations of the increasingly popular notion of differential privacy. We establish a connection between differential private mechanisms and the rate-distortion framework. Additionally, we als... |
1901.09697 | 2966439069 | We propose Bayesian differential privacy, a relaxation of differential privacy for similarly distributed data, that provides sharper privacy guarantees in difficult scenarios, such as deep learning. We also derive a general privacy accounting method for iterative learning algorithms under Bayesian differential privacy ... | Apart from sharp bounds, moments accountant is attractive because it operates within the classical notion of @math -DP. Alternative notions of DP also provide tight composition theorems, along with some other advantages, but to the best of our knowledge, are not broadly used in practice compared to traditional DP (alth... | {
"cite_N": [
"@cite_1"
],
"mid": [
"2963828152"
],
"abstract": [
"With the newly proposed privacy definition of Renyi Differential Privacy (RDP) in (Mironov, 2017), we re-examine the inherent privacy of releasing a single sample from a posterior distribution. We exploit the impact of the prior di... |
1901.09697 | 2966439069 | We propose Bayesian differential privacy, a relaxation of differential privacy for similarly distributed data, that provides sharper privacy guarantees in difficult scenarios, such as deep learning. We also derive a general privacy accounting method for iterative learning algorithms under Bayesian differential privacy ... | We evaluate our method on two popular classes of learning algorithms: deep neural networks and variational inference (VI). Privacy-preserving deep learning is now extensively studied, and is frequently used in combination with moments accountant @cite_27 @cite_25 @cite_28 , which makes it a perfect setting for comparis... | {
"cite_N": [
"@cite_7",
"@cite_8",
"@cite_28",
"@cite_3",
"@cite_27",
"@cite_2",
"@cite_25"
],
"mid": [
"1886087434",
"1553111043",
"2785361959",
"2539938672",
"2473418344",
"2608609295",
"2950602864"
],
"abstract": [
"We consider the problem of Bay... |
1901.09681 | 2913663242 | We study the problem of identifying different behaviors occurring in different parts of a large heterogenous network. We zoom in to the network using lenses of different sizes to capture the local structure of the network. These network signatures are then weighted to provide a set of predicted labels for every node. W... | The idea of using the image embedding of the adjacency matrix as a feature was first introduced in @cite_6 . Based on this idea, authors in @cite_7 showed with great success that parent networks of tiny subgraphs (as small as 8 nodes) can be identified. They also used Caffe @cite_8 to show that the structured image emb... | {
"cite_N": [
"@cite_7",
"@cite_6",
"@cite_8"
],
"mid": [
"2786815098",
"2554819193",
"2950094539"
],
"abstract": [
"We propose a novel subgraph image representation for classification of network fragments with the target being their parent networks. The graph image representation ... |
1901.09608 | 2913504012 | Sensors are routinely mounted on robots to acquire various forms of measurements in spatiotemporal fields. Locating features within these fields and reconstruction (mapping) of the dense fields can be challenging in resource-constrained situations, such as when trying to locate the source of a gas leak from a small num... | Early work around autonomous sensing of physical phenomena involved ground-based mobile robots @cite_26 @cite_30 @cite_40 . More recently, with the emergence of reasonably robust Unmanned Aerial Vehicle (UAV) platforms, often referred to as drones, they are being used as sensing platforms with benefits in terms of spee... | {
"cite_N": [
"@cite_30",
"@cite_26",
"@cite_19",
"@cite_40",
"@cite_10",
"@cite_25"
],
"mid": [
"",
"2131262329",
"2615505359",
"2007236156",
"",
"2553399731"
],
"abstract": [
"",
"In this paper we introduce a statistical method to build two-dimensional... |
1901.09608 | 2913504012 | Sensors are routinely mounted on robots to acquire various forms of measurements in spatiotemporal fields. Locating features within these fields and reconstruction (mapping) of the dense fields can be challenging in resource-constrained situations, such as when trying to locate the source of a gas leak from a small num... | Models provide numerous advantages in machine learning @cite_11 , enabling inferences from limited data, and in planning @cite_29 , enabling counter-factual reasoning @cite_12 and guided search. However, defining the structure of models in a way that leads to efficient inference while maintaining fidelity to complex ar... | {
"cite_N": [
"@cite_29",
"@cite_12",
"@cite_11"
],
"mid": [
"2479648088",
"2210428523",
"2137077888"
],
"abstract": [
"Autonomous AI systems need complex computational techniques for planning and performing actions. Planning and acting require significant deliberation because an i... |
1901.09608 | 2913504012 | Sensors are routinely mounted on robots to acquire various forms of measurements in spatiotemporal fields. Locating features within these fields and reconstruction (mapping) of the dense fields can be challenging in resource-constrained situations, such as when trying to locate the source of a gas leak from a small num... | The phenomena we consider in this paper involve gas flows. There is a long tradition of modelling such flows, including efficient computational methods aimed at graphics and animation applications @cite_42 . The development of efficient solvers is also driven in the engineering community by the need to simulate phenome... | {
"cite_N": [
"@cite_18",
"@cite_42",
"@cite_31",
"@cite_21"
],
"mid": [
"2115513040",
"2889337996",
"2112961542",
"2653160509"
],
"abstract": [
"In fluid simulation, enforcing incompressibility is crucial for realism; it is also computationally expensive. Recent work has i... |
1901.09608 | 2913504012 | Sensors are routinely mounted on robots to acquire various forms of measurements in spatiotemporal fields. Locating features within these fields and reconstruction (mapping) of the dense fields can be challenging in resource-constrained situations, such as when trying to locate the source of a gas leak from a small num... | In this paper, we utilise a reasonably accurate simulation of the phenomenon @cite_23 but exploit simplifications inherent to the problem, such as that the dispersion process can be modelled on the 2-d plane We observe that our approach is invariant to some degree of (small) noise, i.e., the situation of plain fields a... | {
"cite_N": [
"@cite_36",
"@cite_23"
],
"mid": [
"2099631309",
"24392662"
],
"abstract": [
"In this paper, a fuzzy model is suggested for the prediction of wind speed and the produced electrical power at a wind park. The model is trained using a genetic algorithm-based learning scheme. The... |
1901.09608 | 2913504012 | Sensors are routinely mounted on robots to acquire various forms of measurements in spatiotemporal fields. Locating features within these fields and reconstruction (mapping) of the dense fields can be challenging in resource-constrained situations, such as when trying to locate the source of a gas leak from a small num... | Another aspect of active sensing is the method for collecting samples so as to maximise a notion of information gain. While the underlying exploration-exploitation tradeoffs can be posed formally in decision theoretic terms, most practical techniques tend to be myopic in their operation. Gaussian processes le2009trajec... | {
"cite_N": [
"@cite_41",
"@cite_38",
"@cite_14",
"@cite_13"
],
"mid": [
"2900643257",
"2887271066",
"",
"2343661315"
],
"abstract": [
"Gaining information about an unknown gas source is a task of great importance with applications in several areas including: responding to ... |
1901.09839 | 2911535260 | While the success of deep neural networks (DNNs) is well-established across a variety of domains, our ability to explain and interpret these methods is limited. Unlike previously proposed local methods which try to explain particular classification decisions, we focus on global interpretability and ask a universally ap... | One viable approach for achieving global interpretability is to train more conventional statistical methods to mimic the predictive behavior of a DNN. This imitation model is then retrospectively used to explain the predictions that a DNN would make. For example, using a decision tree @cite_18 or falling rule list @cit... | {
"cite_N": [
"@cite_24",
"@cite_18",
"@cite_28",
"@cite_23"
],
"mid": [
"2964134873",
"2769421449",
"2050063758",
""
],
"abstract": [
"Falling rule lists are classification models consisting of an ordered list of if-then rules, where (i) the order of rules determines which... |
1901.09590 | 2914592219 | Knowledge graphs are structured representations of real world facts. However, they typically contain only a small subset of all possible facts. Link prediction is a task of inferring missing facts based on existing ones. We propose TuckER, a relatively simple but powerful linear model based on Tucker decomposition of t... | RESCAL An early linear model, RESCAL @cite_18 , optimizes a scoring function containing a bilinear product between vector embeddings for each subject and object entity and a full rank matrix for each relation. Although a very expressive and powerful model, RESCAL is prone to overfitting due to its large number of param... | {
"cite_N": [
"@cite_18"
],
"mid": [
"205829674"
],
"abstract": [
"Relational learning is becoming increasingly important in many areas of application. Here, we present a novel approach to relational learning based on the factorization of a three-way tensor. We show that unlike other tensor approa... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.