aid stringlengths 9 15 | mid stringlengths 7 10 | abstract stringlengths 78 2.56k | related_work stringlengths 92 1.77k | ref_abstract dict |
|---|---|---|---|---|
1908.01549 | 2964399413 | Code obfuscation is a major tool for protecting software intellectual property from attacks such as reverse engineering or code tampering. Yet, recently proposed (automated) attacks based on Dynamic Symbolic Execution (DSE) shows very promising results, hence threatening software integrity. Current defenses are not ful... | We have already discsussed obfuscation, symbolic execution and symbolic deobfuscation at length throughout the paper, including successful applications of DSE-related techniques to deobfuscation @cite_17 @cite_48 @cite_44 @cite_46 . In addition, @cite_2 give an exhaustive survey about program analysis-based deobfuscati... | {
"cite_N": [
"@cite_30",
"@cite_48",
"@cite_44",
"@cite_2",
"@cite_46",
"@cite_17"
],
"mid": [
"2138788987",
"1538186256",
"2806540472",
"2314464932",
"2708742135",
"2049629426"
],
"abstract": [
"Dynamic taint analysis and forward symbolic execution are qui... |
1908.01549 | 2964399413 | Code obfuscation is a major tool for protecting software intellectual property from attacks such as reverse engineering or code tampering. Yet, recently proposed (automated) attacks based on Dynamic Symbolic Execution (DSE) shows very promising results, hence threatening software integrity. Current defenses are not ful... | @cite_34 describe, in the setting of automatic testing, the three major weaknesses of DSE: , and . Cadar @cite_19 shows that compiler optimizations can sensibly alter the performance of a symbolic analyzer like , confirming the folklore knowledge that strong enough compiler optimizations resemble code obfuscations. Tha... | {
"cite_N": [
"@cite_19",
"@cite_34"
],
"mid": [
"1989310296",
"1979345446"
],
"abstract": [
"Semantics-preserving program transformations, such as refactorings and optimisations, can have a significant impact on the effectiveness of symbolic execution testing and analysis. Furthermore, se... |
1908.01549 | 2964399413 | Code obfuscation is a major tool for protecting software intellectual property from attacks such as reverse engineering or code tampering. Yet, recently proposed (automated) attacks based on Dynamic Symbolic Execution (DSE) shows very promising results, hence threatening software integrity. Current defenses are not ful... | Most anti-DSE techniques target the constraint solving engine through hard-to-solve predicates. The impact on symbolic deobfuscation through the complexification of constraints has been studied by @cite_11 . @cite_41 propose an obfuscation based on expressions @cite_23 to complexify points-to functions , making it hard... | {
"cite_N": [
"@cite_41",
"@cite_28",
"@cite_23",
"@cite_5",
"@cite_25",
"@cite_11"
],
"mid": [
"2620363344",
"",
"1514687127",
"109909280",
"2963452421",
"2751013686"
],
"abstract": [
"Control flow obfuscation techniques can be used to hinder software rever... |
1908.01549 | 2964399413 | Code obfuscation is a major tool for protecting software intellectual property from attacks such as reverse engineering or code tampering. Yet, recently proposed (automated) attacks based on Dynamic Symbolic Execution (DSE) shows very promising results, hence threatening software integrity. Current defenses are not ful... | Yadegari and Debray @cite_31 describe obfuscations thwarting standard byte-level taint analysis, possibly resulting in missing legitimate paths for DSE engines using taint analysis ( does, and do not). It can be circumvented in the case of taint-based DSE by bit-level tainting @cite_31 . combines this idea with input-d... | {
"cite_N": [
"@cite_31",
"@cite_8",
"@cite_36",
"@cite_29",
"@cite_46",
"@cite_16"
],
"mid": [
"2010417554",
"2873938857",
"2560252021",
"2012737144",
"2708742135",
"1496704300"
],
"abstract": [
"Symbolic and concolic execution nd important applications in ... |
1908.01549 | 2964399413 | Code obfuscation is a major tool for protecting software intellectual property from attacks such as reverse engineering or code tampering. Yet, recently proposed (automated) attacks based on Dynamic Symbolic Execution (DSE) shows very promising results, hence threatening software integrity. Current defenses are not ful... | @cite_36 set the ground for the experimental evaluation of symbolic deobfuscation techniques. Our own experimental evaluation extends and refines their method in several ways: new metrics, different DSE settings, larger examples. @cite_43 propose a mathematically proven obfuscation against Abstract Model Checking attac... | {
"cite_N": [
"@cite_36",
"@cite_43"
],
"mid": [
"2560252021",
"2780799849"
],
"abstract": [
"Code obfuscation is widely used by software developers to protect intellectual property, and malware writers to hamper program analysis. However, there seems to be little work on systematic evalua... |
1908.01449 | 2966779809 | Training deep neural networks typically requires large amounts of labeled data which may be scarce or expensive to obtain for a particular target domain. As an alternative, we can leverage webly-supervised data (i.e. results from a public search engine) which are relatively plentiful but may contain noisy results. In t... | . Previous work using data include @cite_13 @cite_32 @cite_29 @cite_4 . Gan al @cite_7 jointly match images and frames in a pre-processing step before using a classifier while LeadExceed @cite_17 uses multiple steps to filter out noisy images and frames. In contrast, our model does not have pre-processing steps and lea... | {
"cite_N": [
"@cite_4",
"@cite_7",
"@cite_29",
"@cite_32",
"@cite_13",
"@cite_17"
],
"mid": [
"2604113307",
"2514167171",
"2287866429",
"2048261712",
"2226011429",
"2469228190"
],
"abstract": [
"Current action recognition methods heavily rely on trimmed vid... |
1908.01449 | 2966779809 | Training deep neural networks typically requires large amounts of labeled data which may be scarce or expensive to obtain for a particular target domain. As an alternative, we can leverage webly-supervised data (i.e. results from a public search engine) which are relatively plentiful but may contain noisy results. In t... | Li al @cite_21 use web images to perform domain adaptation and learn a video classifier, but they manually filter out irrelevant web images beforehand whereas we incorporate this step into our model. | {
"cite_N": [
"@cite_21"
],
"mid": [
"2742318416"
],
"abstract": [
"Training deep learning based video classifiers for action recognition requires a large amount of labeled videos. The labeling process is labor-intensive and time-consuming. On the other hand, large amount of weakly-labeled images ... |
1908.01449 | 2966779809 | Training deep neural networks typically requires large amounts of labeled data which may be scarce or expensive to obtain for a particular target domain. As an alternative, we can leverage webly-supervised data (i.e. results from a public search engine) which are relatively plentiful but may contain noisy results. In t... | There has also been related work in using attention for weakly-supervised learning. Zhuang al @cite_24 stack noisy web image features together with the assumption that at least one of the images is correctly labeled. They then learn an attention model to focus on the correctly labeled images. UntrimmedNet @cite_4 gener... | {
"cite_N": [
"@cite_24",
"@cite_4"
],
"mid": [
"2949499025",
"2604113307"
],
"abstract": [
"Large-scale datasets have driven the rapid development of deep neural networks for visual recognition. However, annotating a massive dataset is expensive and time-consuming. Web images and their la... |
1908.01449 | 2966779809 | Training deep neural networks typically requires large amounts of labeled data which may be scarce or expensive to obtain for a particular target domain. As an alternative, we can leverage webly-supervised data (i.e. results from a public search engine) which are relatively plentiful but may contain noisy results. In t... | . 3D-CNN video models such as C3D @cite_27 , P3D @cite_3 , I3D @cite_9 , @cite_30 are appealing for video classification since they learn appearance and motion features jointly. I3D @cite_9 uses full 3D filters, while @cite_30 and P3D @cite_3 decompose the spatio-temporal convolution into a spatial convolution followed... | {
"cite_N": [
"@cite_30",
"@cite_9",
"@cite_27",
"@cite_3"
],
"mid": [
"2772114784",
"2619082050",
"2952633803",
"2761659801"
],
"abstract": [
"In this paper we discuss several forms of spatiotemporal convolutions for video analysis and study their effects on action recogni... |
1908.01449 | 2966779809 | Training deep neural networks typically requires large amounts of labeled data which may be scarce or expensive to obtain for a particular target domain. As an alternative, we can leverage webly-supervised data (i.e. results from a public search engine) which are relatively plentiful but may contain noisy results. In t... | . There has been much work in adapting GANs @cite_11 for domain adaptation. Models such as PixelDA @cite_1 learn to generate realistic-looking samples from the source distribution, while others such as DANN @cite_5 learn a domain-invariant feature representation. We adopt this latter approach in our work. Other related... | {
"cite_N": [
"@cite_26",
"@cite_28",
"@cite_1",
"@cite_5",
"@cite_12",
"@cite_11"
],
"mid": [
"2953226914",
"",
"2949212125",
"",
"2949987290",
"2099471712"
],
"abstract": [
"Recent reports suggest that a generic supervised deep CNN model trained on a large... |
1908.01449 | 2966779809 | Training deep neural networks typically requires large amounts of labeled data which may be scarce or expensive to obtain for a particular target domain. As an alternative, we can leverage webly-supervised data (i.e. results from a public search engine) which are relatively plentiful but may contain noisy results. In t... | The main difference between our model and these approaches is that we use data and assume the source and target domains may contain noisy labels, which is a considerably more difficult yet practical scenario. Lastly there is recent work by Zhang al @cite_22 that is similar to our model in that they also have a domain-a... | {
"cite_N": [
"@cite_22"
],
"mid": [
"2795066828"
],
"abstract": [
"This paper proposes an importance weighted adversarial nets-based method for unsupervised domain adaptation, specific for partial domain adaptation where the target domain has less number of classes compared to the source domain. ... |
1908.01441 | 2965360919 | A partial edge drawing (PED) of a graph is a variation of a node-link diagram. PED draws a link, which is a partial visual representation of an edge, and reduces visual clutter of the node-link diagram. However, more time is required to read a PED to infer undrawn parts. The authors propose a morphing edge drawing (MED... | conceived a drawing concept in which only half the links are used to reduce the visual clutter during the development of a tool called SeeNet @cite_13 . Parallel Tagcloud, developed by , adopts a method similar to PED @cite_11 . Although Parallel Tagcloud is an extension of Tag Cloud, it can be regarded as a hierarchic... | {
"cite_N": [
"@cite_13",
"@cite_11"
],
"mid": [
"2059942088",
"2075038621"
],
"abstract": [
"Networks are critical to modern society, and a thorough understanding of how they behave is crucial to their efficient operation. Fortunately, data on networks is plentiful; by visualizing this da... |
1908.01441 | 2965360919 | A partial edge drawing (PED) of a graph is a variation of a node-link diagram. PED draws a link, which is a partial visual representation of an edge, and reduces visual clutter of the node-link diagram. However, more time is required to read a PED to infer undrawn parts. The authors propose a morphing edge drawing (MED... | formalized the PED in their study @cite_6 ; our formalization of PED in is a modified version of their formalization. They added continuity of the omitted parts of links as a condition, we have incorporated this into the formalization herein. Moreover, although they focused on a layout without crossing stubs in PED, th... | {
"cite_N": [
"@cite_9",
"@cite_1",
"@cite_6"
],
"mid": [
"2403365036",
"2168788994",
"160524401"
],
"abstract": [
"",
"We investigate the readability of node-link diagrams for directed graphs when using partially drawn links instead of showing each link explicitly in its full ... |
1908.01441 | 2965360919 | A partial edge drawing (PED) of a graph is a variation of a node-link diagram. PED draws a link, which is a partial visual representation of an edge, and reduces visual clutter of the node-link diagram. However, more time is required to read a PED to infer undrawn parts. The authors propose a morphing edge drawing (MED... | performed a comparison between CED They call it the traditional straight-line model (TRA). and 1 4-SHPED on reading performance of graphs @cite_0 . Although the statistical significance was not shown, from their chart that visualizes the experimental results, in the task of reading graphs (for adjacency check of two no... | {
"cite_N": [
"@cite_0",
"@cite_5",
"@cite_10"
],
"mid": [
"2406810977",
"2567569351",
"2770013200"
],
"abstract": [
"Partial edge drawing PED is a model for a straight-line drawing of a graph, where edges are subdivided into three parts in order to drop the middle part.",
"Par... |
1908.01441 | 2965360919 | A partial edge drawing (PED) of a graph is a variation of a node-link diagram. PED draws a link, which is a partial visual representation of an edge, and reduces visual clutter of the node-link diagram. However, more time is required to read a PED to infer undrawn parts. The authors propose a morphing edge drawing (MED... | avoided using arrows to facilitate grasping high-dimensional transitions in the state transition diagram and proposed moving the dashed pattern with animation @cite_4 . compared the recognition accuracy of graphs in various edge drawing methods, such as tapered links and curved links, including animation @cite_7 . They... | {
"cite_N": [
"@cite_4",
"@cite_12",
"@cite_7"
],
"mid": [
"2117161218",
"2789365861",
"2122675754"
],
"abstract": [
"In this paper, we present a new visual way of exploring state sequences in large observational time-series. A key advantage of our method is that it can directly vi... |
1908.01379 | 2964814808 | Depth acquisition, based on active illumination, is essential for autonomous and robotic navigation. LiDARs (Light Detection And Ranging) with mechanical, fixed, sampling templates are commonly used in today's autonomous vehicles. An emerging technology, based on solid-state depth sensors, with no mechanical parts, all... | Among the unguided methods, some use classical approach @cite_27 @cite_33 , while others rely on more advanced tools such as deep learning @cite_13 @cite_19 . | {
"cite_N": [
"@cite_19",
"@cite_27",
"@cite_13",
"@cite_33"
],
"mid": [
"2750215633",
"2787455114",
"2794558150",
"2593751597"
],
"abstract": [
"In this paper, we consider convolutional neural networks operating on sparse inputs with an application to depth upsampling from... |
1908.01379 | 2964814808 | Depth acquisition, based on active illumination, is essential for autonomous and robotic navigation. LiDARs (Light Detection And Ranging) with mechanical, fixed, sampling templates are commonly used in today's autonomous vehicles. An emerging technology, based on solid-state depth sensors, with no mechanical parts, all... | On the contrary, guided methods exploit the connection between depth maps and their corresponding color image. Earlier methods used traditional image processing tools @cite_11 @cite_15 . Recently, several deep learning-based methods @cite_26 @cite_25 @cite_4 @cite_17 @cite_3 @cite_30 @cite_14 @cite_1 achieved state-of-... | {
"cite_N": [
"@cite_30",
"@cite_26",
"@cite_4",
"@cite_14",
"@cite_1",
"@cite_17",
"@cite_3",
"@cite_15",
"@cite_25",
"@cite_11"
],
"mid": [
"2949661992",
"2796461436",
"2888634437",
"2810837462",
"2760112017",
"2886851716",
"2810477386",
""... |
1908.01379 | 2964814808 | Depth acquisition, based on active illumination, is essential for autonomous and robotic navigation. LiDARs (Light Detection And Ranging) with mechanical, fixed, sampling templates are commonly used in today's autonomous vehicles. An emerging technology, based on solid-state depth sensors, with no mechanical parts, all... | Early guided depth sampling: Despite the intensive development in depth completion, the issue of adaptive sampling is yet little addressed. Only @cite_5 @cite_21 have offered a non-trivial (i.e. uniformly random or grid) sampling pattern as a previous step to depth reconstruction. Both studies selected sampling at loca... | {
"cite_N": [
"@cite_22",
"@cite_7",
"@cite_21",
"@cite_24",
"@cite_5",
"@cite_31"
],
"mid": [
"607933207",
"1973233493",
"2005441409",
"2139242438",
"1967027087",
"2004421506"
],
"abstract": [
"1. Introduction F. Marvasti. 2. An Introduction to Sampling Ana... |
1908.01423 | 2966059979 | Games are often designed to shape player behavior in a desired way; however, it can be unclear how design decisions affect the space of behaviors in a game. Designers usually explore this space through human playtesting, which can be time-consuming and of limited effectiveness in exhausting the space of possible behavi... | Automated game generation researchers have simulated player behavior using search to reach goal states @cite_34 @cite_30 @cite_47 (possibly subject to constraints on the states visited in the reachability check @cite_22 ) and hand-coded heuristics for learnability @cite_5 or design aesthetics such as balance @cite_42 @... | {
"cite_N": [
"@cite_30",
"@cite_35",
"@cite_4",
"@cite_22",
"@cite_7",
"@cite_46",
"@cite_9",
"@cite_42",
"@cite_27",
"@cite_23",
"@cite_5",
"@cite_47",
"@cite_34"
],
"mid": [
"",
"1515504392",
"",
"2395654457",
"",
"1976948167",
"20... |
1908.01423 | 2966059979 | Games are often designed to shape player behavior in a desired way; however, it can be unclear how design decisions affect the space of behaviors in a game. Designers usually explore this space through human playtesting, which can be time-consuming and of limited effectiveness in exhausting the space of possible behavi... | Evaluating potential player behavior for an arbitrary game design is a central concern of general game playing research @cite_41 @cite_8 . Recently, Monte Carlo Tree Search (MCTS) has emerged as a popular technique in general game playing after the successful application of MCTS to the game of Go @cite_10 . Game applic... | {
"cite_N": [
"@cite_38",
"@cite_37",
"@cite_8",
"@cite_41",
"@cite_36",
"@cite_28",
"@cite_21",
"@cite_1",
"@cite_3",
"@cite_24",
"@cite_10",
"@cite_17"
],
"mid": [
"",
"1991331404",
"",
"2099587183",
"2077755260",
"158154636",
"78804240... |
1908.01207 | 2965683718 | Modeling sequential interactions between users and items products is crucial in domains such as e-commerce, social networking, and education. Representation learning presents an attractive opportunity to model the dynamic evolution of users and items, where each user item can be embedded in a Euclidean space and its ev... | Several models have recently been developed that generate embeddings for the nodes (users and items) in temporal networks. CTDNE @cite_15 is a state-of-the-art algorithm that generates embeddings using temporally-increasing random walks, but it generates one final static embedding of the nodes. Similarly, IGE @cite_12 ... | {
"cite_N": [
"@cite_38",
"@cite_31",
"@cite_41",
"@cite_48",
"@cite_46",
"@cite_44",
"@cite_43",
"@cite_40",
"@cite_15",
"@cite_51",
"@cite_12"
],
"mid": [
"2787927827",
"2790197930",
"1917516492",
"2797520986",
"2763783057",
"2804558096",
"... |
1908.01367 | 2965871844 | For ego-motion estimation, the feature representation of the scenes is crucial. Previous methods indicate that both the low-level and semantic feature-based methods can achieve promising results. Therefore, the incorporation of hierarchical feature representation may benefit from both methods. From this perspective, we... | The SLAM field was dominated by feature-based (indirect) methods for a long time. In recent years, a number of different approaches have gained in popularity, such as direct methods and semantic SLAM @cite_10 . Unlike feature-based (indirect) visual odometry methods (e.g., ORB-SLAM @cite_21 ), direct visual odometry me... | {
"cite_N": [
"@cite_21",
"@cite_10"
],
"mid": [
"2535547924",
"2739423245"
],
"abstract": [
"We present ORB-SLAM2, a complete simultaneous localization and mapping (SLAM) system for monocular, stereo and RGB-D cameras, including map reuse, loop closing, and relocalization capabilities. Th... |
1908.01367 | 2965871844 | For ego-motion estimation, the feature representation of the scenes is crucial. Previous methods indicate that both the low-level and semantic feature-based methods can achieve promising results. Therefore, the incorporation of hierarchical feature representation may benefit from both methods. From this perspective, we... | These methods may fail under certain conditions such as low texture, stereo ambiguities and occlusions. The quality of feature points is essential for the feature-based SLAM method. Agrawal al @cite_15 show that features learned from a convolution neural network outperform the traditional hand-designed features. Learni... | {
"cite_N": [
"@cite_15"
],
"mid": [
"2951590555"
],
"abstract": [
"The dominant paradigm for feature learning in computer vision relies on training neural networks for the task of object recognition using millions of hand labelled images. Is it possible to learn useful features for a diverse set ... |
1908.01367 | 2965871844 | For ego-motion estimation, the feature representation of the scenes is crucial. Previous methods indicate that both the low-level and semantic feature-based methods can achieve promising results. Therefore, the incorporation of hierarchical feature representation may benefit from both methods. From this perspective, we... | Zhou al @cite_38 propose a framework that jointly trains the pose estimation and depth prediction networks. In addition to the depth maps and camera poses, the work of Vijayanarasimhan al @cite_6 can predict scene motion from an additional subnetwork. To improve the quality of estimated depth values, several constraint... | {
"cite_N": [
"@cite_38",
"@cite_7",
"@cite_9",
"@cite_29",
"@cite_6"
],
"mid": [
"2609883120",
"",
"2785512290",
"2773227160",
"2608018946"
],
"abstract": [
"We present an unsupervised learning framework for the task of monocular depth and camera motion estimation ... |
1908.01367 | 2965871844 | For ego-motion estimation, the feature representation of the scenes is crucial. Previous methods indicate that both the low-level and semantic feature-based methods can achieve promising results. Therefore, the incorporation of hierarchical feature representation may benefit from both methods. From this perspective, we... | Currently, determining the approach to incorporate the traditional SLAM methods with deep neural networks is still an open problem. The depth prediction neural network can be used in a variety of scenarios, such as keyframe initialization @cite_46 , scale recovery and virtual stereo constraints @cite_20 . CodeSLAM @cit... | {
"cite_N": [
"@cite_46",
"@cite_20",
"@cite_39"
],
"mid": [
"2952280228",
"",
"2795447360"
],
"abstract": [
"Given the recent advances in depth prediction from Convolutional Neural Networks (CNNs), this paper investigates how predicted depth maps from a deep neural network can be ... |
1908.01294 | 2966008508 | A sentence is typically treated as the minimal syntactic unit used for extracting valuable information from a longer piece of text. However, in written Thai, there are no explicit sentence markers. We proposed a deep learning model for the task of sentence segmentation that includes three main contributions. First, we ... | Focusing only on textual features, there are two main approaches, namely, word sequence tagging and machine translation. For the machine translation approach, punctuation is treated as just another type of token that needs to be recovered and included in the output. The methods in @cite_23 @cite_11 @cite_24 restore pun... | {
"cite_N": [
"@cite_24",
"@cite_23",
"@cite_11"
],
"mid": [
"2899902398",
"2406343628",
"2403334028"
],
"abstract": [
"Inserting proper punctuation into Automatic Speech Recognizer(ASR) transcription is a challenging and promising task in real-time Spoken Language Translation(SLT)... |
1908.01293 | 2965124455 | Visual localization is the problem of estimating a camera within a scene and a key component in computer vision applications such as self-driving cars and Mixed Reality. State-of-the-art approaches for accurate visual localization use scene-specific representations, resulting in the overhead of constructing these model... | Machine learning in visual localization. Retrieval methods @cite_62 @cite_39 @cite_3 have benefitted greatly from deep learning architectures. Several works have been proposed to directly learn the 2D-3D matching function @cite_37 @cite_0 @cite_57 @cite_1 @cite_28 @cite_79 @cite_45 While traditionally a dense 3D scene ... | {
"cite_N": [
"@cite_37",
"@cite_62",
"@cite_28",
"@cite_1",
"@cite_3",
"@cite_39",
"@cite_0",
"@cite_57",
"@cite_79",
"@cite_45",
"@cite_84"
],
"mid": [
"2950681848",
"2951019013",
"2081605477",
"",
"2144990732",
"",
"2768185495",
"25931... |
1908.01293 | 2965124455 | Visual localization is the problem of estimating a camera within a scene and a key component in computer vision applications such as self-driving cars and Mixed Reality. State-of-the-art approaches for accurate visual localization use scene-specific representations, resulting in the overhead of constructing these model... | Learning absolute pose estimation. Recent works have proposed to learn the complete localization pipeline, by either modeling localization as a classification problem @cite_18 or learning to directly regress the absolute pose of a query image @cite_14 @cite_76 @cite_22 @cite_54 @cite_20 @cite_91 . These methods typical... | {
"cite_N": [
"@cite_18",
"@cite_14",
"@cite_22",
"@cite_91",
"@cite_54",
"@cite_76",
"@cite_20"
],
"mid": [
"2284646714",
"2951336016",
"2605111497",
"2584731199",
"",
"2279895976",
"2771385090"
],
"abstract": [
"Is it possible to determine the loca... |
1908.01293 | 2965124455 | Visual localization is the problem of estimating a camera within a scene and a key component in computer vision applications such as self-driving cars and Mixed Reality. State-of-the-art approaches for accurate visual localization use scene-specific representations, resulting in the overhead of constructing these model... | Learning relative pose estimation. The authors of DeMoN @cite_55 propose a CNN that jointly predicts a depth map for one image and the relative pose to a second image. They require depth maps for training, whereas our approach does not. An unsupervised variant of DeMoN that is trained purely on a stream of images by us... | {
"cite_N": [
"@cite_48",
"@cite_55"
],
"mid": [
"2609883120",
"2561074213"
],
"abstract": [
"We present an unsupervised learning framework for the task of monocular depth and camera motion estimation from unstructured video sequences. In common with recent work [10, 14, 16], we use an end... |
1908.01007 | 2966576392 | Training deep reinforcement learning agents complex behaviors in 3D virtual environments requires significant computational resources. This is especially true in environments with high degrees of aliasing, where many states share nearly identical visual features. Minecraft is an exemplar of such an environment. We hypo... | Reinforcement Learning is proven to be effective at learning agent control policies in 2D and 3D game environments including Atari @cite_19 , navigation in Minecraft environments @cite_22 @cite_13 @cite_2 , Doom @cite_16 , Quake II @cite_4 , DOTA 2, and StarCraft II. Most game environments are designed to have low perc... | {
"cite_N": [
"@cite_4",
"@cite_22",
"@cite_9",
"@cite_19",
"@cite_2",
"@cite_16",
"@cite_13"
],
"mid": [
"2810602713",
"2944804155",
"2021171258",
"2121092017",
"2164424353",
"2362143032",
"2950560044"
],
"abstract": [
"Reinforcement learning (RL) h... |
1908.01007 | 2966576392 | Training deep reinforcement learning agents complex behaviors in 3D virtual environments requires significant computational resources. This is especially true in environments with high degrees of aliasing, where many states share nearly identical visual features. Minecraft is an exemplar of such an environment. We hypo... | Human feedback in IML can take different forms. Learning from Demonstration (LfD) allows humans to directly provide examples of proper behavior @cite_0 . From demonstrations, an agent can learn the policy directly, learn to explore more effectively @cite_18 , or learn a reward function from which to reconstruct a polic... | {
"cite_N": [
"@cite_18",
"@cite_14",
"@cite_21",
"@cite_0",
"@cite_11"
],
"mid": [
"2098441518",
"1999874108",
"2626804490",
"1986014385",
"2963571918"
],
"abstract": [
"A long term goal of Interactive Reinforcement Learning is to incorporate nonexpert human feedba... |
1908.01007 | 2966576392 | Training deep reinforcement learning agents complex behaviors in 3D virtual environments requires significant computational resources. This is especially true in environments with high degrees of aliasing, where many states share nearly identical visual features. Minecraft is an exemplar of such an environment. We hypo... | We build upon a rich set of research exploring human feedback in reinforcement learning @cite_18 @cite_11 @cite_3 @cite_6 @cite_10 @cite_24 . A majority of this work assesses the algorithms in 2D grid-world environments where the agent’s @math location is a given feature. Some do address 3D @cite_17 , more simplistic 3... | {
"cite_N": [
"@cite_18",
"@cite_7",
"@cite_17",
"@cite_6",
"@cite_3",
"@cite_24",
"@cite_10",
"@cite_12",
"@cite_11"
],
"mid": [
"2098441518",
"2754794180",
"2785379783",
"2116372557",
"121023703",
"1589747210",
"2151074445",
"2299638491",
"... |
1908.01007 | 2966576392 | Training deep reinforcement learning agents complex behaviors in 3D virtual environments requires significant computational resources. This is especially true in environments with high degrees of aliasing, where many states share nearly identical visual features. Minecraft is an exemplar of such an environment. We hypo... | We build off two LfA algorithms in our exploration of perceptual aliasing in 3D virtual environments. The first, Feedback Arbitration @cite_7 is based on a standard deep reinforcement learning technique called Deep @math Networks (DQN). Deep @math Networks @cite_8 use a convolutional neural network to learn visual feat... | {
"cite_N": [
"@cite_7",
"@cite_8"
],
"mid": [
"2754794180",
"2145339207"
],
"abstract": [
"We describe a method to use discrete human feedback to enhance the performance of deep learning agents in virtual three-dimensional environments by extending deep-reinforcement learning to model the... |
1908.01007 | 2966576392 | Training deep reinforcement learning agents complex behaviors in 3D virtual environments requires significant computational resources. This is especially true in environments with high degrees of aliasing, where many states share nearly identical visual features. Minecraft is an exemplar of such an environment. We hypo... | The second, the Newtonian Action Advice algorithm @cite_11 is built on tabular @math -learning, where a table of @math -values is stored for every state and action. found that human trainers perceived a Newtonian Action Advice agent as more intelligent, more transparent, more performant and less frustrating than a stan... | {
"cite_N": [
"@cite_11"
],
"mid": [
"2963571918"
],
"abstract": [
"A goal of Interactive Machine Learning is to enable people without specialized training to teach agents how to perform tasks. Many of the existing algorithms that learn from human instructions are evaluated using simulated feedbac... |
1908.01010 | 2965781583 | Aleatoric uncertainty is an intrinsic property of ill-posed inverse and imaging problems. Its quantification is vital for assessing the reliability of relevant point estimates. In this paper, we propose an efficient framework for quantifying aleatoric uncertainty for deep residual learning and showcase its significant ... | Random generators are an inherent part of deep generative models, e.g. VAEs and GANs. Since generative distributions are unconditional, samples are directly generated in a latent space and transformed into the target space. In conditional cases, the latent distributions are conditioned on the input variable which lives... | {
"cite_N": [
"@cite_9",
"@cite_27",
"@cite_59"
],
"mid": [
"2807861409",
"2942471690",
"2952232639"
],
"abstract": [
"Many real-world vision problems suffer from inherent ambiguities. In clinical applications for example, it might not be clear from a CT scan alone which particular... |
1908.01010 | 2965781583 | Aleatoric uncertainty is an intrinsic property of ill-posed inverse and imaging problems. Its quantification is vital for assessing the reliability of relevant point estimates. In this paper, we propose an efficient framework for quantifying aleatoric uncertainty for deep residual learning and showcase its significant ... | In many VAE CVAE applications and their variants, a multiplier @math to the KL term is often introduced. The work @cite_54 identified the KL vanishing problem and proposed a sigmoid annealing scheme, and the work @cite_37 suggested a linear annealing scheme. By annealing the multiplier @math , the objective function co... | {
"cite_N": [
"@cite_55",
"@cite_54",
"@cite_37",
"@cite_49"
],
"mid": [
"",
"2210838531",
"2807753879",
"2753738274"
],
"abstract": [
"",
"The standard recurrent neural network language model (RNNLM) generates sentences one word at a time and does not work from an expl... |
1908.00953 | 2966834907 | Most of the current active queue management (AQM) designs have major issues including severe hardship of being tuned for highly fluctuated cellular access link bandwidths. Consequently, most of the cellular network providers either give up using AQMs or use conservative offline configurations for them. However, these c... | Explicit congestion notification (ECN) is another technique used to do the buffer management indirectly. In other words, ECN-based approaches, use modified switches to tag certain packets and explicitly notify the senders about congestion in the network. Later, senders need to adjust their sending rates to reduce queue... | {
"cite_N": [
"@cite_28",
"@cite_19"
],
"mid": [
"2745034467",
"2769057555"
],
"abstract": [
"We present our experience with QUIC, an encrypted, multiplexed, and low-latency transport protocol designed from the ground up to improve transport performance for HTTPS traffic and to enable rapi... |
1908.00672 | 2965514655 | We show that existing upsampling operators can be unified with the notion of the index function. This notion is inspired by an observation in the decoding process of deep image matting where indices-guided unpooling can recover boundary details much better than other upsampling operators such as bilinear interpolation.... | Upsampling is an essential stage for almost all dense prediction tasks. It has been intensively studied about (decoding). The operator, also known as transposed convolution, was initially used in @cite_15 to visualize convolutional activations and latter introduced to semantic segmentation @cite_40 . To avoid checkerbo... | {
"cite_N": [
"@cite_28",
"@cite_9",
"@cite_29",
"@cite_24",
"@cite_40",
"@cite_5",
"@cite_15"
],
"mid": [
"",
"2964309882",
"",
"2963881378",
"1903029394",
"2476548250",
"1849277567"
],
"abstract": [
"",
"Spatial pyramid pooling module or encode... |
1908.00672 | 2965514655 | We show that existing upsampling operators can be unified with the notion of the index function. This notion is inspired by an observation in the decoding process of deep image matting where indices-guided unpooling can recover boundary details much better than other upsampling operators such as bilinear interpolation.... | Our work is primarily inspired by the unpooling operator @cite_24 . We remark that, it is important to keep the spatial information before loss of such information occurred in feature map downsampling, and more importantly, to use stored information during upsampling. Unpooling shows a simple and effective case of doin... | {
"cite_N": [
"@cite_24"
],
"mid": [
"2963881378"
],
"abstract": [
"We present a novel and practical deep fully convolutional neural network architecture for semantic pixel-wise segmentation termed SegNet. This core trainable segmentation engine consists of an encoder network, a corresponding deco... |
1908.00672 | 2965514655 | We show that existing upsampling operators can be unified with the notion of the index function. This notion is inspired by an observation in the decoding process of deep image matting where indices-guided unpooling can recover boundary details much better than other upsampling operators such as bilinear interpolation.... | In the past decades, image matting methods have been extensively studied from a low-level view @cite_13 @cite_26 @cite_27 @cite_25 @cite_46 @cite_49 @cite_38 @cite_19 @cite_39 ; and particularly, they have been designed to solve the matting equation. Despite being theoretically elegant, these methods heavily rely on th... | {
"cite_N": [
"@cite_38",
"@cite_26",
"@cite_39",
"@cite_19",
"@cite_27",
"@cite_49",
"@cite_46",
"@cite_13",
"@cite_25"
],
"mid": [
"",
"",
"",
"2035773017",
"2098554565",
"2077604257",
"1560942455",
"2963304750",
"2103917701"
],
"abstra... |
1908.00672 | 2965514655 | We show that existing upsampling operators can be unified with the notion of the index function. This notion is inspired by an observation in the decoding process of deep image matting where indices-guided unpooling can recover boundary details much better than other upsampling operators such as bilinear interpolation.... | With the tremendous success of deep CNNs in high-level vision tasks @cite_31 @cite_0 @cite_40 , deep matting methods are emerging. Some initial attempts appeared in @cite_11 and @cite_22 , where classic matting approaches, such as closed-form matting @cite_19 and KNN matting @cite_26 , are still used as the backends in... | {
"cite_N": [
"@cite_26",
"@cite_4",
"@cite_22",
"@cite_48",
"@cite_1",
"@cite_0",
"@cite_19",
"@cite_40",
"@cite_24",
"@cite_2",
"@cite_31",
"@cite_11"
],
"mid": [
"",
"2807926603",
"2518810941",
"2962826264",
"2891201469",
"2163605009",
... |
1908.00672 | 2965514655 | We show that existing upsampling operators can be unified with the notion of the index function. This notion is inspired by an observation in the decoding process of deep image matting where indices-guided unpooling can recover boundary details much better than other upsampling operators such as bilinear interpolation.... | SegNet is effective in matting but also computation-expensive and memory-inefficient. For instance, the inference can only be executed on CPU when testing high-resolution images, which is practically unattractive. We show that, with our proposed IndexNet, even a lightweight backbone such as MobileNetv2-based model can ... | {
"cite_N": [
"@cite_2"
],
"mid": [
"2604469346"
],
"abstract": [
"Image matting is a fundamental computer vision problem and has many applications. Previous algorithms have poor performance when an image has similar foreground and background colors or complicated textures. The main reasons are pr... |
1908.00677 | 2964947356 | We study the Riemannian quantiative isoperimetric inequality. We show that direct analogue of the Euclidean quantitative isoperimetric inequality is--in general--false on a closed Riemannian manifold. In spite of this, we show that the inequality is true generically. Moreover, we show that a modified (but sharp) versio... | As mentioned above, there has been a lot of recent work on quantitative stability not just for the isoperimetric inequality but also for many other geometric (e.g. Brunn-Minkowski @cite_3 ), spectral (e.g. Faber-Krahn @cite_30 ), and functional (e.g. Sobolev @cite_33 ) inequalities. We defer to the recent survey of Fus... | {
"cite_N": [
"@cite_30",
"@cite_35",
"@cite_33",
"@cite_3",
"@cite_0",
"@cite_23",
"@cite_13"
],
"mid": [
"1560993323",
"2888452402",
"2913618394",
"2964176678",
"",
"",
"1935847907"
],
"abstract": [
"The classical Faber-Krahn inequality asserts tha... |
1908.00677 | 2964947356 | We study the Riemannian quantiative isoperimetric inequality. We show that direct analogue of the Euclidean quantitative isoperimetric inequality is--in general--false on a closed Riemannian manifold. In spite of this, we show that the inequality is true generically. Moreover, we show that a modified (but sharp) versio... | In the anisotropic setting, optimal transport techniques have been used with great success (see, e.g. @cite_41 ). However, usually convexity of the extremizers is required (e.g. to guarantee the necessary regularity on the transport map). Other techniques, such as the selection principle, often require understanding th... | {
"cite_N": [
"@cite_41",
"@cite_4"
],
"mid": [
"2963747906",
"2075584439"
],
"abstract": [
"By using optimal mass transportation and a quantitative Holder inequality, we provide estimates for the Borell–Brascamp–Lieb deficit on complete Riemannian manifolds. Accordingly, equality cases in... |
1908.00310 | 2966086679 | This paper considers the problem of estimating a power-law degree distribution of an undirected network. Although power-law degree distributions are ubiquitous in nature, the widely used parametric methods for estimating them (e.g. linear regression on double-logarithmic axes, maximum likelihood estimation with uniform... | It has been shown in the literature that maximum likelihood methods are more suitable for estimating power-law degree distributions of the form in Eq. compared to alternative methods @cite_34 @cite_36 @cite_53 . Further, exploiting the friendship paradox (Theorem ) has shown to be effective in empirical estimation of h... | {
"cite_N": [
"@cite_36",
"@cite_53",
"@cite_34",
"@cite_50"
],
"mid": [
"2140849606",
"2000042664",
"2027963417",
"1493228775"
],
"abstract": [
"Distributions following a power-law are an ubiquitous phenomenon. Methods for determining the exponent of a power-law tail by gr... |
1908.00524 | 2965027008 | Cameras and 2D laser scanners, in combination, are able to provide low-cost, light-weight and accurate solutions, which make their fusion well-suited for many robot navigation tasks. However, correct data fusion depends on precise calibration of the rigid body transform between the sensors. In this paper we present the... | The application of Deep Learning techniques has presented impressive results recently in tasks such as object detection and classification by the use of camera images. To obtain these results, a large amount of datasets has become available in the last years. In the context of intelligent vehicles, interesting work has... | {
"cite_N": [
"@cite_9",
"@cite_15",
"@cite_18",
"@cite_8"
],
"mid": [
"2869375357",
"2963647876",
"2896832493",
"2963914175"
],
"abstract": [
"We present research using the latest reinforcement learning algorithm for end-to-end driving without any mediated perception (obje... |
1908.00524 | 2965027008 | Cameras and 2D laser scanners, in combination, are able to provide low-cost, light-weight and accurate solutions, which make their fusion well-suited for many robot navigation tasks. However, correct data fusion depends on precise calibration of the rigid body transform between the sensors. In this paper we present the... | At the same time, localization, which is still a very challenging problem for robotic systems, is not yet well explored by deep learning methods. The most common approaches are based on the use of camera images for odometry estimation. They are inspired by the classic methods for Visual Odometry (VO) @cite_2 @cite_17 ,... | {
"cite_N": [
"@cite_14",
"@cite_11",
"@cite_2",
"@cite_13",
"@cite_17"
],
"mid": [
"2598706937",
"2964314455",
"1612997784",
"2200124539",
"1970504153"
],
"abstract": [
"This paper studies monocular visual odometry (VO) problem. Most of existing VO algorithms are d... |
1908.00524 | 2965027008 | Cameras and 2D laser scanners, in combination, are able to provide low-cost, light-weight and accurate solutions, which make their fusion well-suited for many robot navigation tasks. However, correct data fusion depends on precise calibration of the rigid body transform between the sensors. In this paper we present the... | Laser scanners are also popular for classic pose estimation because of its accuracy. Classic approaches for this problem consist in trying to match two point clouds and estimate the transformation between them, this solution is known as Iterative Closest Point (ICP) @cite_16 . Since then, more robust and complex algori... | {
"cite_N": [
"@cite_16",
"@cite_12"
],
"mid": [
"2083624955",
"2296228853"
],
"abstract": [
"This paper describes a general purpose, representation independent method for the accurate and computationally efficient registration of 3-D shapes including free-form curves and surfaces. The met... |
1908.00524 | 2965027008 | Cameras and 2D laser scanners, in combination, are able to provide low-cost, light-weight and accurate solutions, which make their fusion well-suited for many robot navigation tasks. However, correct data fusion depends on precise calibration of the rigid body transform between the sensors. In this paper we present the... | The application of deep learning techniques for this purpose using laser scanners is still considered as a new challenge and only few papers have addressed it. @cite_21 were the first to propose to apply 3D laser scanner data in CNNs to estimate odometry. Their approach provides a reasonable estimation of odometry, how... | {
"cite_N": [
"@cite_21",
"@cite_10",
"@cite_12"
],
"mid": [
"2774351931",
"",
"2296228853"
],
"abstract": [
"Although 2D LiDAR based Simultaneous Localization and Mapping (SLAM) is a relatively mature topic nowadays, the loop closure problem remains challenging due to the lack of ... |
1908.00524 | 2965027008 | Cameras and 2D laser scanners, in combination, are able to provide low-cost, light-weight and accurate solutions, which make their fusion well-suited for many robot navigation tasks. However, correct data fusion depends on precise calibration of the rigid body transform between the sensors. In this paper we present the... | The use of 2D laser scanners instead of 3D laser scanners could reduce considerably the price of future intelligent vehicles and the need of high computational resources. The authors of this work presented in @cite_5 a solution that relies only on a 2D laser scanner for odometry estimation using Recurrent Convolutional... | {
"cite_N": [
"@cite_5"
],
"mid": [
"2932490017"
],
"abstract": [
"The use of 2D laser scanners is attractive for the autonomous driving industry because of its accuracy, light-weight and low-cost. However, since only a 2D slice of the surrounding environment is detected at each scan, it is a chal... |
1908.00524 | 2965027008 | Cameras and 2D laser scanners, in combination, are able to provide low-cost, light-weight and accurate solutions, which make their fusion well-suited for many robot navigation tasks. However, correct data fusion depends on precise calibration of the rigid body transform between the sensors. In this paper we present the... | We present the first deep learning method for odometry estimation based on the fusion of a 2D laser scanner and a camera. The proposed network is able to provide a real-time solution that overcome the difficulties enconter by one sensor alone. We also present how to transform the odometry regression problem into a into... | {
"cite_N": [
"@cite_22"
],
"mid": [
"2150066425"
],
"abstract": [
"Today, visual recognition systems are still rarely employed in robotics applications. Perhaps one of the main reasons for this is the lack of demanding benchmarks that mimic such scenarios. In this paper, we take advantage of our ... |
1908.00432 | 2966501858 | Payment channels allow transactions between participants of the blockchain to be executed securely off-chain, and thus provide a promising solution for the scalability problem of popular blockchains. We study the online network design problem for payment channels, assuming a central coordinator. We focus on a single ch... | Additionally, the design of payment networks with fees from the viewpoint of a payment service provider who wants to maximise the profit is discussed in @cite_15 . In the contrary to this work where we assume constant fee for every transaction, in @cite_15 , each channel requires a different fee, much like the tolls on... | {
"cite_N": [
"@cite_15"
],
"mid": [
"2890103577"
],
"abstract": [
"Payment channels are the most prominent solution to the blockchain scalability problem. We introduce the problem of network design with fees for payment channels from the perspective of a Payment Service Provider (PSP). Given a se... |
1908.00439 | 2965874802 | In this paper, we tackle the problem of 3D human shape estimation from single RGB images. While the recent progress in convolutional neural networks has allowed impressive results for 3D human pose estimation, estimating the full 3D shape of a person is still an open issue. Model-based approaches can output precise mes... | 3D human body shape from images. Most existing methods for body shape estimation from single images rely on a parametric model of the human body whose pose and shape parameters are optimized to match image evidence @cite_5 @cite_21 @cite_18 @cite_35 . This optimization process is usually initialised with an estimate of... | {
"cite_N": [
"@cite_35",
"@cite_18",
"@cite_37",
"@cite_38",
"@cite_7",
"@cite_28",
"@cite_9",
"@cite_21",
"@cite_2",
"@cite_5",
"@cite_10"
],
"mid": [
"2951940669",
"2573098616",
"",
"",
"2778680124",
"2562703451",
"2333544554",
"254517... |
1908.00439 | 2965874802 | In this paper, we tackle the problem of 3D human shape estimation from single RGB images. While the recent progress in convolutional neural networks has allowed impressive results for 3D human pose estimation, estimating the full 3D shape of a person is still an open issue. Model-based approaches can output precise mes... | 3D human datasets. Current approaches for human 3D pose estimation are built on deep architectures trained and evaluated on large datasets acquired in controlled environments with Motion Capture systems @cite_45 @cite_44 @cite_14 . However, while the typology of human poses on these datasets captures the space of human... | {
"cite_N": [
"@cite_18",
"@cite_14",
"@cite_42",
"@cite_6",
"@cite_39",
"@cite_44",
"@cite_45"
],
"mid": [
"2573098616",
"2099333815",
"2895748257",
"2888934629",
"2605947573",
"2101032778",
""
],
"abstract": [
"3D models provide a common ground for... |
1908.00113 | 2965823887 | Physical phenomena in science and engineering are frequently modeled using scalar fields. In scalar field topology, graph-based topological descriptors such as merge trees, contour trees, and Reeb graphs are commonly used to characterize topological changes in the (sub)level sets of scalar fields. One of the biggest ch... | We review the most relevant literature on uncertainty visualization of scalar fields, with a focus on graph-based topological descriptors and topological features. Our work is primarily concerned with structure-wise data uncertainty of scalar fields; for vector and tensor fields, see an overview in @cite_31 . | {
"cite_N": [
"@cite_31"
],
"mid": [
"2475195788"
],
"abstract": [
"This paper presents the state of the art in the area of topology-based visualization. It describes the process and results of an extensive annotation for generating a definition and terminology for the field. The terminology enabl... |
1908.00113 | 2965823887 | Physical phenomena in science and engineering are frequently modeled using scalar fields. In scalar field topology, graph-based topological descriptors such as merge trees, contour trees, and Reeb graphs are commonly used to characterize topological changes in the (sub)level sets of scalar fields. One of the biggest ch... | Graph-based topological descriptors. Graph-based topological descriptors include merge trees @cite_20 (also known as barrier trees @cite_37 or join trees @cite_17 ), contour trees @cite_17 , Reeb graphs @cite_16 , mapper graphs @cite_12 , and joint contour nets @cite_27 . These descriptors are graph-based representatio... | {
"cite_N": [
"@cite_37",
"@cite_27",
"@cite_16",
"@cite_20",
"@cite_12",
"@cite_17"
],
"mid": [
"1979184481",
"2031993264",
"",
"",
"1607787316",
"2001088180"
],
"abstract": [
"The heights of energy barriers separating two (macro-)states are useful for esti... |
1908.00113 | 2965823887 | Physical phenomena in science and engineering are frequently modeled using scalar fields. In scalar field topology, graph-based topological descriptors such as merge trees, contour trees, and Reeb graphs are commonly used to characterize topological changes in the (sub)level sets of scalar fields. One of the biggest ch... | Since a contour tree of a function @math can be constructed by carefully combining the merge trees of @math and @math in linear time @cite_17 , merge tree visualization shares the same design space as that of a contour tree. Contour trees are often visualized with node-link diagrams in two or three dimensions @cite_43 ... | {
"cite_N": [
"@cite_8",
"@cite_60",
"@cite_24",
"@cite_43",
"@cite_79",
"@cite_59",
"@cite_71",
"@cite_17"
],
"mid": [
"2083586903",
"2135609792",
"2070784608",
"2160747282",
"",
"2011658854",
"1584564047",
"2001088180"
],
"abstract": [
"Top... |
1908.00113 | 2965823887 | Physical phenomena in science and engineering are frequently modeled using scalar fields. In scalar field topology, graph-based topological descriptors such as merge trees, contour trees, and Reeb graphs are commonly used to characterize topological changes in the (sub)level sets of scalar fields. One of the biggest ch... | Mihai and Westermann @cite_55 measure the likelihood of the occurrence of critical points with respect to both the positions and types of the critical points. Specifically, when the data uncertainty is described by a Gaussian distribution, confidence intervals are derived for the gradient and the determinant and trace ... | {
"cite_N": [
"@cite_55",
"@cite_75"
],
"mid": [
"1988628849",
"1604082823"
],
"abstract": [
"Abstract In scalar fields, critical points (points with vanishing derivatives) are important indicators of the topology of iso-contours. When the data values are affected by uncertainty, the locat... |
1908.00113 | 2965823887 | Physical phenomena in science and engineering are frequently modeled using scalar fields. In scalar field topology, graph-based topological descriptors such as merge trees, contour trees, and Reeb graphs are commonly used to characterize topological changes in the (sub)level sets of scalar fields. One of the biggest ch... | To visualize the effect of uncertainty on contours, envelopes within a gridded domain are extracted to indicate in which volume the contour will lie (with a certain confidence) @cite_11 @cite_26 . Uncertainty associated with a contour can also be rendered via animation @cite_33 or as a collection of points where each p... | {
"cite_N": [
"@cite_38",
"@cite_26",
"@cite_33",
"@cite_41",
"@cite_36",
"@cite_32",
"@cite_63",
"@cite_5",
"@cite_11"
],
"mid": [
"2095233635",
"1972641242",
"2038628506",
"2114084204",
"2109657057",
"2020226548",
"2014256895",
"2031619085",
... |
1908.00113 | 2965823887 | Physical phenomena in science and engineering are frequently modeled using scalar fields. In scalar field topology, graph-based topological descriptors such as merge trees, contour trees, and Reeb graphs are commonly used to characterize topological changes in the (sub)level sets of scalar fields. One of the biggest ch... | Finally, from an algorithmic perspective, probabilistic marching cubes @cite_2 and positionally uncertain iso-contours @cite_38 study the uncertainties inherent in computing the visual representations. | {
"cite_N": [
"@cite_38",
"@cite_2"
],
"mid": [
"2095233635",
"2049885731"
],
"abstract": [
"Uncertainty is ubiquitous in science, engineering and medicine. Drawing conclusions from uncertain data is the normal case, not an exception. While the field of statistical graphics is well establi... |
1908.00113 | 2965823887 | Physical phenomena in science and engineering are frequently modeled using scalar fields. In scalar field topology, graph-based topological descriptors such as merge trees, contour trees, and Reeb graphs are commonly used to characterize topological changes in the (sub)level sets of scalar fields. One of the biggest ch... | Uncertainty visualization of graph-based decriptors. have introduced CandidTree, which merges two trees into one and visualizes both location and subtree structural uncertainty @cite_7 . have developed an interactive visualization tool that uses contour trees as abstract data representations to explore data-level uncer... | {
"cite_N": [
"@cite_18",
"@cite_54",
"@cite_13",
"@cite_7"
],
"mid": [
"2051308568",
"2296659714",
"68571785",
"2128347847"
],
"abstract": [
"Uncertainty is a common and crucial issue in scientific data. The exploration and analysis of three-dimensional (3D) and large two-... |
1908.00113 | 2965823887 | Physical phenomena in science and engineering are frequently modeled using scalar fields. In scalar field topology, graph-based topological descriptors such as merge trees, contour trees, and Reeb graphs are commonly used to characterize topological changes in the (sub)level sets of scalar fields. One of the biggest ch... | Distances between topological structures. Recently, many metrics have been proposed for merge trees, often by way of a restriction from a metric on the more general Reeb graph @cite_20 @cite_80 @cite_1 @cite_39 @cite_34 @cite_4 @cite_50 @cite_61 @cite_64 . In this paper, we focus on the interleaving distance for labele... | {
"cite_N": [
"@cite_61",
"@cite_64",
"@cite_4",
"@cite_22",
"@cite_21",
"@cite_1",
"@cite_39",
"@cite_45",
"@cite_50",
"@cite_80",
"@cite_34",
"@cite_25",
"@cite_20"
],
"mid": [
"2895105248",
"",
"2335518360",
"2806011050",
"2148561979",
... |
1908.00113 | 2965823887 | Physical phenomena in science and engineering are frequently modeled using scalar fields. In scalar field topology, graph-based topological descriptors such as merge trees, contour trees, and Reeb graphs are commonly used to characterize topological changes in the (sub)level sets of scalar fields. One of the biggest ch... | The application of the interleaving distance to labeled merge trees can also be viewed as an interpretation of a metric for phylogenetic trees @cite_68 . Our framework differs from the previous work as it relies on a clean and simple metric-space view of input trees that is equipped with geodesics, as well as easy-to-i... | {
"cite_N": [
"@cite_68"
],
"mid": [
"2108380304"
],
"abstract": [
"Background Phylogenetic tree comparison metrics are an important tool in the study of evolution, and hence the definition of such metrics is an interesting problem in phylogenetics. In a paper in Taxon fifty years ago, Sokal and R... |
1908.00205 | 2944546684 | Abstract The purpose of network representation is to learn a set of latent features by obtaining community information from network structures to provide knowledge for machine learning tasks. Recent research has driven significant progress in network representation by employing random walks as the network sampling stra... | The previously used per-node partition function @cite_26 is expensive to compute, especially for large information networks. To overcome this disadvantage, a series of sampling strategies have been proposed @cite_44 @cite_23 to analyze the statistics within local structures, e.g., communities and sub-networks. These ap... | {
"cite_N": [
"@cite_26",
"@cite_22",
"@cite_8",
"@cite_9",
"@cite_44",
"@cite_24",
"@cite_23",
"@cite_34",
"@cite_20"
],
"mid": [
"2161506909",
"1971955426",
"2242161203",
"2099192586",
"2106315062",
"2083666679",
"2405982180",
"2579372251",
... |
1908.00205 | 2944546684 | Abstract The purpose of network representation is to learn a set of latent features by obtaining community information from network structures to provide knowledge for machine learning tasks. Recent research has driven significant progress in network representation by employing random walks as the network sampling stra... | DeepWalk @cite_1 trains a neural language model on the random walks generated by the network structure. After denoting a random walk that starts from a root node, DeepWalk slides a window and maps the central node to its representation. Hierarchical Softmax factors out the probability distributions corresponding to the... | {
"cite_N": [
"@cite_19",
"@cite_18",
"@cite_29",
"@cite_1"
],
"mid": [
"1888005072",
"2962756421",
"2607500032",
"2154851992"
],
"abstract": [
"This paper studies the problem of embedding very large information networks into low-dimensional vector spaces, which is useful i... |
1908.00205 | 2944546684 | Abstract The purpose of network representation is to learn a set of latent features by obtaining community information from network structures to provide knowledge for machine learning tasks. Recent research has driven significant progress in network representation by employing random walks as the network sampling stra... | All the previous works based on random walk to sample networks into a steam of nodes are under a common assumption of power-law distribution. The power-law distribution exists widely in real-world networks. It is a special degree distribution that follows @math , where @math is a node degree and @math is a positive con... | {
"cite_N": [
"@cite_42",
"@cite_1",
"@cite_6",
"@cite_31",
"@cite_13",
"@cite_11"
],
"mid": [
"1527463082",
"2154851992",
"2950627632",
"1995213405",
"1994687138",
""
],
"abstract": [
"Many communication and social networks have power-law link distributions... |
1908.00205 | 2944546684 | Abstract The purpose of network representation is to learn a set of latent features by obtaining community information from network structures to provide knowledge for machine learning tasks. Recent research has driven significant progress in network representation by employing random walks as the network sampling stra... | In this paper, we propose CDNR which employs biased random walk sampling strategies to learn network structures based on previous works. However, CDNR is different from the deep transfer learning approaches for cross-domain graph-structured data, i.e., context enhanced inductive representation @cite_12 , intrinsic geom... | {
"cite_N": [
"@cite_25",
"@cite_32",
"@cite_12"
],
"mid": [
"2899831870",
"2604545383",
"2962767366"
],
"abstract": [
"This paper presents a general inductive graph representation learning framework called DeepGL for learning deep node and edge features that generalize across-netw... |
1908.00041 | 2964608077 | Vector spherical harmonics on @math have wide applications in geophysics, quantum mechanics and astrophysics. In the representation of a tangent field, one needs to evaluate the expansion and the Fourier coefficients of vector spherical harmonics. In this paper, we develop fast algorithms (FaVeST) for vector spherical ... | In contrast, fast transforms for vector spherical harmonics receive less attention. To the best of our knowledge, there are no existing fast algorithms for the forward and adjoint vector spherical harmonic transforms. @cite_58 simply use FFTs to speed up their algorithms for solving Navier-Stokes PDEs on the unit spher... | {
"cite_N": [
"@cite_53",
"@cite_58"
],
"mid": [
"2767502902",
"1977548994"
],
"abstract": [
"We present in this paper a spectrally accurate numerical method for computing the spherical vector spherical harmonic expansion of a function vector field with given (elemental) nodal values on a ... |
1908.00220 | 2964850415 | To interpret the meanings of colors in visualizations of categorical information, people must determine how distinct colors correspond to different concepts. This process is easier when assignments between colors and concepts in visualizations match people's expectations, making color palettes semantically interpretabl... | Several factors are relevant when designing color palettes for visualizing categorical information. First and foremost, colors that represent different categories must appear different; @cite_6 @cite_29 @cite_5 @cite_31 . Other considerations include selecting colors that have distinct names @cite_57 , are aestheticall... | {
"cite_N": [
"@cite_30",
"@cite_18",
"@cite_29",
"@cite_6",
"@cite_56",
"@cite_57",
"@cite_19",
"@cite_40",
"@cite_5",
"@cite_31",
"@cite_20"
],
"mid": [
"1977400967",
"2611374796",
"",
"2128852435",
"",
"2137167595",
"",
"2788435232",
... |
1908.00112 | 2966640145 | We investigate the problem of cost-optimal planning in ASP. Current ASP planners can be trivially extended to a cost-optimal one by adding weak constraints, but only for a given makespan (number of steps). It is desirable to have a planner that guarantees global optimality. In this paper, we present two approaches to a... | Our 2-threaded solver algorithm is inspired by the approach of @cite_19 . The better performance of our planner, besides solver technologies, seems partly due to the grounding size and search space pruning under stable model semantics as commented in Section , and partly due to less clustered encoding in ASP than in SA... | {
"cite_N": [
"@cite_19"
],
"mid": [
"2152383085"
],
"abstract": [
"We consider the problem of computing optimal plans for propositional planning problems with action costs. In the spirit of leveraging advances in general-purpose automated reasoning for that setting, we develop an approach that op... |
1908.00112 | 2966640145 | We investigate the problem of cost-optimal planning in ASP. Current ASP planners can be trivially extended to a cost-optimal one by adding weak constraints, but only for a given makespan (number of steps). It is desirable to have a planner that guarantees global optimality. In this paper, we present two approaches to a... | Delete-free planning has been investigated as a stand-alone topic, including a CP solution @cite_20 . To the best of our knowledge, our modeling of delete-free planning as a graph problem is original, and it leads to a five-line ASP program which does everything (cf. Appendix C). | {
"cite_N": [
"@cite_20"
],
"mid": [
"2200283827"
],
"abstract": [
"This paper deals with planning domains that appear in computer games, especially when modeling intelligent virtual agents. Some of these domains contain only actions with no negative effects and are thus treated as easy from the p... |
1908.00112 | 2966640145 | We investigate the problem of cost-optimal planning in ASP. Current ASP planners can be trivially extended to a cost-optimal one by adding weak constraints, but only for a given makespan (number of steps). It is desirable to have a planner that guarantees global optimality. In this paper, we present two approaches to a... | The Madagascar Planner is a family of efficient implementations of the SAT based techniques for planning. The main idea is, instead of using the standard decision heuristics such as VSIDS, planning-specific variable selection heuristics are applied @cite_7 . One would expect that the same idea can work for ASP-based pl... | {
"cite_N": [
"@cite_7"
],
"mid": [
"2048644716"
],
"abstract": [
"Reduction to SAT is a very successful approach to solving hard combinatorial problems in Artificial Intelligence and computer science in general. Most commonly, problem instances reduced to SAT are solved with a general-purpose SAT... |
1908.00112 | 2966640145 | We investigate the problem of cost-optimal planning in ASP. Current ASP planners can be trivially extended to a cost-optimal one by adding weak constraints, but only for a given makespan (number of steps). It is desirable to have a planner that guarantees global optimality. In this paper, we present two approaches to a... | Cost-optimal planners can also be built on the @math platform. An @math planner based on greedy selection of a subset of heuristics for guiding @math search @cite_13 has made to the top tier in IPC-2018. @math planning can be encoded in SAT and ASP as well, but the most critical component, the selection algorithm, need... | {
"cite_N": [
"@cite_13"
],
"mid": [
"2574105631"
],
"abstract": [
"In this paper we present greedy methods for selecting a subset of heuristic functions for guiding A* search. Our methods are able to optimize various objective functions while selecting a subset from a pool of up to thousands of h... |
1908.00112 | 2966640145 | We investigate the problem of cost-optimal planning in ASP. Current ASP planners can be trivially extended to a cost-optimal one by adding weak constraints, but only for a given makespan (number of steps). It is desirable to have a planner that guarantees global optimality. In this paper, we present two approaches to a... | Stepless planning is a brand new approach to logic-based planning and brings with it a lot of unknowns and potentials for future directions. One issue is that the lack of any notion of simultaneity makes certain standard optimizations difficult, such as incorporating mutex constraints and supporting conditional-effects... | {
"cite_N": [
"@cite_1"
],
"mid": [
"2208493917"
],
"abstract": [
"We present the first effective SAT heuristics for planning with expressive planning languages such as ADL. Recently, SAT heuristics for STRIPS planning have been introduced. In this work we show that the basic ideas in the heuristi... |
1908.00112 | 2966640145 | We investigate the problem of cost-optimal planning in ASP. Current ASP planners can be trivially extended to a cost-optimal one by adding weak constraints, but only for a given makespan (number of steps). It is desirable to have a planner that guarantees global optimality. In this paper, we present two approaches to a... | Finally, property directed reachability (PDR), a promising method for deciding reachability in symbolically represented transition systems, which was originally conceived as a model checking algorithm for hardware circuits, has recently been related to planning @cite_0 . The relationship with our stepless planner deser... | {
"cite_N": [
"@cite_0"
],
"mid": [
"1540013686"
],
"abstract": [
"Property Directed Reachability (PDR) is a very promising recent method for deciding reachability in symbolically represented transition systems. While originally conceived as a model checking algorithm for hardware circuits, it has... |
1902.00623 | 2950876045 | Cross-modal similarity search is a problem about designing a search system supporting querying across content modalities, e.g., using an image to search for texts or using a text to search for images. This paper presents a compact coding solution for efficient search, with a focus on the quantization approach which has... | Cross-view hashing @cite_28 defines the distance between documents in the Hamming space by considering the hash codes of all the modalities, and aligns it with the given similarity. Multi-view spectral hashing @cite_36 adopts a similar formulation but with a different optimization algorithm. These methods usually also ... | {
"cite_N": [
"@cite_28",
"@cite_10",
"@cite_36",
"@cite_20"
],
"mid": [
"",
"2162670057",
"189447628",
"2025432722"
],
"abstract": [
"",
"Similarity search applications with a large amount of text and image data demands an efficient and effective solution. One useful s... |
1902.00623 | 2950876045 | Cross-modal similarity search is a problem about designing a search system supporting querying across content modalities, e.g., using an image to search for texts or using a text to search for images. This paper presents a compact coding solution for efficient search, with a focus on the quantization approach which has... | The relation is often used to learn a unified hash code, into which a hash function is learnt for each modality to map the feature. For example, Latent semantic sparse hashing @cite_24 applies the sign function on the joint space projected from the latent semantic representation learnt for each modality. Collective mat... | {
"cite_N": [
"@cite_30",
"@cite_35",
"@cite_38",
"@cite_32",
"@cite_3",
"@cite_24",
"@cite_16",
"@cite_10"
],
"mid": [
"2213253086",
"2159373756",
"154472438",
"2251084241",
"1976258951",
"2086958058",
"1922199343",
"2162670057"
],
"abstract": [... |
1902.00623 | 2950876045 | Cross-modal similarity search is a problem about designing a search system supporting querying across content modalities, e.g., using an image to search for texts or using a text to search for images. This paper presents a compact coding solution for efficient search, with a focus on the quantization approach which has... | Recently, a few techniques based on quantization are developed for cross-modal search. Quantized correlation hashing @cite_33 combines the hash function learning with the quantization by minimizing the inter-modality similarity disagreement as well as the binary quantization simultaneously. Compositional correlation qu... | {
"cite_N": [
"@cite_37",
"@cite_33"
],
"mid": [
"1780444626",
"2267050401"
],
"abstract": [
"Efficient similarity retrieval from large-scale multimodal database is pervasive in current search systems with the big data tidal wave. To support queries across content modalities, the system sh... |
1902.00438 | 2918540534 | The use of background knowledge remains largely unexploited in many text classification tasks. In this work, we explore word taxonomies as means for constructing new semantic features, which may improve the performance and robustness of the learned classifiers. We propose tax2vec, a parallel algorithm for constructing ... | Document classification is highly dependent on . In simple bag-of-words representations, the frequency (or a similar weight such as term frequency-inverse document frequency---tf-idf) of each word or @math -gram is considered as a separate feature. More advanced representations group words with similar meaning together... | {
"cite_N": [
"@cite_41",
"@cite_59",
"@cite_57",
"@cite_12"
],
"mid": [
"2058013187",
"2153579005",
"2019731274",
"1880262756"
],
"abstract": [
"Latent semantic analysis (LSA) approximates human understanding of relations between word and passage meanings in a wide variety... |
1902.00438 | 2918540534 | The use of background knowledge remains largely unexploited in many text classification tasks. In this work, we explore word taxonomies as means for constructing new semantic features, which may improve the performance and robustness of the learned classifiers. We propose tax2vec, a parallel algorithm for constructing ... | In practical applications, features are constructed from various data sources, including texts @cite_27 , graphs @cite_9 , audio recordings and similar data @cite_43 . With the increasing computational power at one's disposal, automated feature construction methods are becoming prevalent. Here, the idea is that given s... | {
"cite_N": [
"@cite_9",
"@cite_27",
"@cite_43"
],
"mid": [
"2905345073",
"2250086655",
""
],
"abstract": [
"Abstract Networked data is data composed of network objects and links. Network objects are characterized by high dimensional attributes and by links indicating the relations... |
1902.00438 | 2918540534 | The use of background knowledge remains largely unexploited in many text classification tasks. In this work, we explore word taxonomies as means for constructing new semantic features, which may improve the performance and robustness of the learned classifiers. We propose tax2vec, a parallel algorithm for constructing ... | Feature selection thus filters out the (unnecessary) features, with the aim of yielding a compact, information-rich representation of the unstructured input. There exist many approaches to feature selection. They can be based on the individual feature's information content, correlation, significance etc. @cite_64 . Fea... | {
"cite_N": [
"@cite_46",
"@cite_64"
],
"mid": [
"1493357981",
"2167101736"
],
"abstract": [
"We summarise various ways of performing dimensionality reduction on high-dimensional microarray data. Many different feature selection and feature extraction methods exist and they are being widel... |
1902.00438 | 2918540534 | The use of background knowledge remains largely unexploited in many text classification tasks. In this work, we explore word taxonomies as means for constructing new semantic features, which may improve the performance and robustness of the learned classifiers. We propose tax2vec, a parallel algorithm for constructing ... | In this section we discuss briefly the works that influenced the development of the proposed approach. One of the most elegant ways to learn from graphs is by transforming them into propositional tables, which are a suitable input for many down-stream learning algorithms. Recent attempts to vectorization of graphs incl... | {
"cite_N": [
"@cite_30",
"@cite_18",
"@cite_26",
"@cite_53",
"@cite_24",
"@cite_5"
],
"mid": [
"2743104969",
"2962756421",
"2964015378",
"2777416523",
"2607500032",
"2962767366"
],
"abstract": [
"We study the problem of representation learning in heterogene... |
1902.00369 | 2912289145 | Accurate segmentation of brain tissue in magnetic resonance images (MRI) is a difficult task due to different types of brain abnormalities. In this paper, we review the deformation method focus on the construction of diffeomorphisms, address clearly a new formation of the deformation problem for moving domains, and we ... | This is known as the deformation method. A series of applications were made including adaptive moving grid @cite_7 , and steady euler flow calculations @cite_5 .In addition,the dynamic version of the deformation method on fixed domains was developed in @cite_7 based on solving Poisson's equations. In 2004,The Least-Squ... | {
"cite_N": [
"@cite_0",
"@cite_5",
"@cite_7"
],
"mid": [
"",
"2028949045",
"1969724971"
],
"abstract": [
"",
"An adaptive remeshing procedure based on a cell volume deformation method is presented. Starting with an initial grid, this method offers direct cell volume control th... |
1902.00369 | 2912289145 | Accurate segmentation of brain tissue in magnetic resonance images (MRI) is a difficult task due to different types of brain abnormalities. In this paper, we review the deformation method focus on the construction of diffeomorphisms, address clearly a new formation of the deformation problem for moving domains, and we ... | The success of the deformation method of grid generation actually relies on the div-curl system, which shows the great importance of the curl vector ( @math ). Hence, in the paper of 2016 @cite_15 , Xi Chen proposed another completely new approach by directly applying JD(the Jacobian determinant @math ) and CV(the curl... | {
"cite_N": [
"@cite_15"
],
"mid": [
"1437816530"
],
"abstract": [
"Adaptive grid generation is an active research topic for numer- ical solution of differential equations. In this paper, we propose a variational method which generates transformations with prescribed Jacobian determinant and curl.... |
1902.00563 | 2914133091 | Keeping the electricity production in balance with the actual demand is becoming a difficult and expensive task in spite of an involvement of experienced human operators. This is due to the increasing complexity of the electric power grid system with the intermittent renewable production as one of the contributors. A b... | The first publication @cite_4 discusses and shows the limitations and insufficiency of richly applied but basic forecasting techniques such as ARIMA and exponential smoothing, because of the non-periodic, non-stationary and noisy character of imbalance time-series. Instead, the contemporary state-of-the-art artificial ... | {
"cite_N": [
"@cite_4"
],
"mid": [
"2158441668"
],
"abstract": [
"Forecasting in power systems has been made considerably more complex by the introduction of competitive electricity markets. Furthermore, new variables need to be predicted by various market participants. This paper shows how a new... |
1902.00489 | 2911903816 | Recent approaches to English-language sentence compression rely on parallel corpora consisting of sentence-compression pairs. However, a sentence may be shortened in many different ways, which each might be suited to the needs of a particular application. Therefore, in this work, we collect and model crowdsourced judge... | Researchers have been studying extractive sentence compression for nearly two decades @cite_24 @cite_37 @cite_7 . Recent approaches are often based on a large compression corpus, https: github.com google-research-datasets sentence-compression which was automatically constructed by using news headlines to identify gold ... | {
"cite_N": [
"@cite_37",
"@cite_7",
"@cite_6",
"@cite_24",
"@cite_43",
"@cite_45"
],
"mid": [
"2115322217",
"2251654079",
"",
"2118119027",
"2251656952",
""
],
"abstract": [
"Sentence compression holds promise for many applications ranging from summarizatio... |
1902.00489 | 2911903816 | Recent approaches to English-language sentence compression rely on parallel corpora consisting of sentence-compression pairs. However, a sentence may be shortened in many different ways, which each might be suited to the needs of a particular application. Therefore, in this work, we collect and model crowdsourced judge... | However, because a sentence may be compressed in many ways (Table ), this work introduces human acceptability judgements as a new and more flexible form of supervision for the sentence compression task. Our approach is thus closely connected to research which seeks to model human judgements of the well-formedness of a ... | {
"cite_N": [
"@cite_4",
"@cite_28",
"@cite_40",
"@cite_15",
"@cite_16"
],
"mid": [
"2592205058",
"",
"2251930319",
"2531882892",
"2806120502"
],
"abstract": [
"Previous investigations into the validity of acceptability judgment data have focused almost exclusively ... |
1902.00505 | 2913527169 | This paper proposes a novel algorithm which learns a formal regular grammar from real-world continuous data, such as videos or other streaming data. Learning latent terminals, non-terminals, and productions rules directly from streaming data allows the construction of a generative model capturing sequential structures ... | Chomsky grammars @cite_2 @cite_3 are designed to represent functional linguistic relationships. They have found wide applications in defining programming languages, natural language understanding, and understanding of images and videos . | {
"cite_N": [
"@cite_3",
"@cite_2"
],
"mid": [
"2040713190",
"2124479173"
],
"abstract": [
"A grammar can be regarded as a device that enumerates the sentences of a language. We study a sequence of restrictions that limit grammars first to Turing machines, then to two types of system from ... |
1902.00617 | 2949537635 | In this paper, we address the problem of searching for semantically similar images from a large database. We present a compact coding approach, supervised quantization. Our approach simultaneously learns feature selection that linearly transforms the database points into a low-dimensional discriminative subspace, and q... | Pairwise similarity preserving hashing aligns the similarity over each pair of items computed in the hash codes with the semantic similarity in various manners. Representative algorithms include LDA Hashing @cite_23 , minimal loss hashing @cite_44 , binary reconstructive embedding @cite_27 , supervised hashing with ker... | {
"cite_N": [
"@cite_36",
"@cite_44",
"@cite_40",
"@cite_27",
"@cite_23",
"@cite_51",
"@cite_11"
],
"mid": [
"2086504823",
"2221852422",
"2143321506",
"2164338181",
"2134514757",
"1992371516",
"1956333070"
],
"abstract": [
"For many computer vision a... |
1902.00617 | 2949537635 | In this paper, we address the problem of searching for semantically similar images from a large database. We present a compact coding approach, supervised quantization. Our approach simultaneously learns feature selection that linearly transforms the database points into a low-dimensional discriminative subspace, and q... | Multiwise similarity preserving hashing formulates the problem by maximizing the agreement of the similarity orders over more than two items between the input space and the coding space. The representative algorithms include order preserving hashing @cite_1 , which directly aligns the rank orders computed from the inpu... | {
"cite_N": [
"@cite_30",
"@cite_38",
"@cite_1",
"@cite_3"
],
"mid": [
"2113307832",
"1923967535",
"2089632823",
"2126210882"
],
"abstract": [
"Motivated by large-scale multimedia applications we propose to learn mappings from high-dimensional data to binary codes that pres... |
1902.00617 | 2949537635 | In this paper, we address the problem of searching for semantically similar images from a large database. We present a compact coding approach, supervised quantization. Our approach simultaneously learns feature selection that linearly transforms the database points into a low-dimensional discriminative subspace, and q... | The recently-developed supervised discrete hashing (SDH) algorithm @cite_25 formulates the problem using the rule that the classification performance over the learned binary codes is as good as possible. This rule seems inferior compared with pairwise and multiwise similarity preserving, but yields superior search perf... | {
"cite_N": [
"@cite_25"
],
"mid": [
"1910300841"
],
"abstract": [
"Recently, learning based hashing techniques have attracted broad research interests because they can support efficient storage and retrieval for high-dimensional data such as images, videos, documents, etc. However, a major diffic... |
1902.00617 | 2949537635 | In this paper, we address the problem of searching for semantically similar images from a large database. We present a compact coding approach, supervised quantization. Our approach simultaneously learns feature selection that linearly transforms the database points into a low-dimensional discriminative subspace, and q... | Our approach is a supervised version of quantization. The quantizer we adopt is composite quantization @cite_15 , which is shown to be a generalized version of product quantization @cite_39 and cartesian k-means @cite_34 , and achieves better performance. Rather than performing the quantization in the input space, our ... | {
"cite_N": [
"@cite_15",
"@cite_34",
"@cite_39"
],
"mid": [
"138284169",
"",
"2124509324"
],
"abstract": [
"This paper presents a novel compact coding approach, composite quantization, for approximate nearest neighbor search. The idea is to use the composition of several elements ... |
1902.00491 | 2914691682 | We present DANTE, a novel method for training neural networks using the alternating minimization principle. DANTE provides an alternate perspective to traditional gradient-based backpropagation techniques commonly used to train deep networks. It utilizes an adaptation of quasi-convexity to cast training a neural networ... | In recent years, Taylor proposed a method to train neural networks using the Alternating Direction Method of Multipliers (ADMM) and Bregman iterations @cite_2 . The focus of this method, however, was on scaling the training of neural networks to a distributed setting on multiple cores across a computing cluster. Jaderb... | {
"cite_N": [
"@cite_11",
"@cite_25",
"@cite_2"
],
"mid": [
"2516591743",
"2809670082",
""
],
"abstract": [
"Training directed neural networks typically requires forward-propagating data through a computation graph, followed by backpropagating error signal, to produce weight update... |
1902.00491 | 2914691682 | We present DANTE, a novel method for training neural networks using the alternating minimization principle. DANTE provides an alternate perspective to traditional gradient-based backpropagation techniques commonly used to train deep networks. It utilizes an adaptation of quasi-convexity to cast training a neural networ... | In our work, we focus on an entirely new approach to training neural networks using alternating optimization and quasi-convexity (different from the abovementioned methods), and show that this approach shows promising results on a range of datasets. Although alternating minimization has found much appeal in areas such ... | {
"cite_N": [
"@cite_22"
],
"mid": [
"2952489973"
],
"abstract": [
"Alternating minimization represents a widely applicable and empirically successful approach for finding low-rank matrices that best fit the given data. For example, for the problem of low-rank matrix completion, this method is bel... |
1902.00491 | 2914691682 | We present DANTE, a novel method for training neural networks using the alternating minimization principle. DANTE provides an alternate perspective to traditional gradient-based backpropagation techniques commonly used to train deep networks. It utilizes an adaptation of quasi-convexity to cast training a neural networ... | Some other efforts have also explored target propagation based methods, such as in @cite_20 , Difference Target Propagation @cite_19 and target propagation in a Bayesian setting @cite_12 . There are also efforts that use random feedback weights such as feedback-alignment @cite_13 and direct indirect feedback-alignment ... | {
"cite_N": [
"@cite_14",
"@cite_19",
"@cite_20",
"@cite_13",
"@cite_12"
],
"mid": [
"2964115671",
"",
"1606458877",
"1931823119",
"1520121841"
],
"abstract": [
"Artificial neural networks are most commonly trained with the back-propagation algorithm, where the grad... |
1907.13615 | 2965843368 | Three-dimensional human body models are widely used in the analysis of human pose and motion. Existing models, however, are learned from minimally-clothed humans and thus do not capture the complexity of dressed humans in common images and videos. To address this, we learn a generative 3D mesh model of clothing from 3D... | Models for clothing and clothed humans. There is a large literature on clothing simulation @cite_12 , which is beyond our scope. Traditional physics simulation requires significant artist or designer involvement and is not practical in the inner loop of a deep learning framework. The most relevant work has focused on c... | {
"cite_N": [
"@cite_38",
"@cite_14",
"@cite_9",
"@cite_52",
"@cite_44",
"@cite_13",
"@cite_12"
],
"mid": [
"1985351089",
"2920928264",
"1978216348",
"2962831356",
"2071839134",
"1967494143",
"2084554572"
],
"abstract": [
"We present a perceptual con... |
1907.13615 | 2965843368 | Three-dimensional human body models are widely used in the analysis of human pose and motion. Existing models, however, are learned from minimally-clothed humans and thus do not capture the complexity of dressed humans in common images and videos. To address this, we learn a generative 3D mesh model of clothing from 3D... | The first problem with real scans is to estimate the shape and pose clothing, which is required to model how garments deviate from the body. Typically, a single shape is optimized over multiple poses to fit inside multiple clothing silhouettes @cite_5 , or a sequence of 3D scans @cite_65 @cite_31 . Here, we use a simil... | {
"cite_N": [
"@cite_26",
"@cite_41",
"@cite_48",
"@cite_55",
"@cite_42",
"@cite_65",
"@cite_21",
"@cite_40",
"@cite_63",
"@cite_5",
"@cite_31",
"@cite_73"
],
"mid": [
"2921745007",
"2887358179",
"1938204631",
"",
"2082145490",
"2596210417",
... |
1907.13615 | 2965843368 | Three-dimensional human body models are widely used in the analysis of human pose and motion. Existing models, however, are learned from minimally-clothed humans and thus do not capture the complexity of dressed humans in common images and videos. To address this, we learn a generative 3D mesh model of clothing from 3D... | Few clothing models, learned from real data, have shown that they generalize to new poses. Neophytou and Hilton @cite_8 learn a layered garment model on top of SCAPE @cite_53 from dynamic sequences, but generalization to novel poses is not demonstrated. Yang al @cite_34 train a neural network to a PCA-based representat... | {
"cite_N": [
"@cite_33",
"@cite_8",
"@cite_53",
"@cite_0",
"@cite_34"
],
"mid": [
"2109683418",
"2071663264",
"1989191365",
"2886416285",
"2887586359"
],
"abstract": [
"We present a new performance capture approach that incorporates a physically-based cloth model t... |
1907.13615 | 2965843368 | Three-dimensional human body models are widely used in the analysis of human pose and motion. Existing models, however, are learned from minimally-clothed humans and thus do not capture the complexity of dressed humans in common images and videos. To address this, we learn a generative 3D mesh model of clothing from 3D... | Generative models for 3D meshes. Generative models for 3D shapes are usually based on PCA @cite_35 or its robust versions @cite_43 . Alternately, deep learning methods such as Variational Autoencoders (VAE) @cite_54 and Generative Adversarial Networks (GAN) @cite_66 have shown state of the art results in generating 2D ... | {
"cite_N": [
"@cite_35",
"@cite_67",
"@cite_22",
"@cite_7",
"@cite_60",
"@cite_54",
"@cite_32",
"@cite_56",
"@cite_39",
"@cite_43",
"@cite_24",
"@cite_72",
"@cite_45",
"@cite_49",
"@cite_66"
],
"mid": [
"2089468765",
"2606840594",
"279456144... |
1907.13615 | 2965843368 | Three-dimensional human body models are widely used in the analysis of human pose and motion. Existing models, however, are learned from minimally-clothed humans and thus do not capture the complexity of dressed humans in common images and videos. To address this, we learn a generative 3D mesh model of clothing from 3D... | To deal with irregular structures like graphs, Bruna al @cite_61 introduce graph convolutions. Follow-up work @cite_50 @cite_28 extends these graph convolutions, which have been successfully used @cite_18 @cite_19 to learn representations defined on meshes. Verma al @cite_19 use feature-steered graph-convolutions for 3... | {
"cite_N": [
"@cite_61",
"@cite_18",
"@cite_28",
"@cite_36",
"@cite_70",
"@cite_19",
"@cite_50",
"@cite_68"
],
"mid": [
"1662382123",
"2883221003",
"2519887557",
"2235901111",
"",
"",
"2964321699",
"2963197375"
],
"abstract": [
"Convolutiona... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.