aid stringlengths 9 15 | mid stringlengths 7 10 | abstract stringlengths 78 2.56k | related_work stringlengths 92 1.77k | ref_abstract dict |
|---|---|---|---|---|
1901.00811 | 2907904605 | Programming a robot to deal with open-ended tasks remains a challenge, in particular if the robot has to manipulate objects. Launching, grasping, pushing or any other object interaction can be simulated but the corresponding models are not reversible and the robot behavior thus cannot be directly deduced. These behavio... | Several methods have been put forward to make robot autonomously explore their environment and learn new skills to reach various goals. An important part of this literature treats this as a control problem, where the system learns an inverse model (or multiple inverse models) to generate the motor commands in order to ... | {
"cite_N": [
"@cite_9",
"@cite_16",
"@cite_23"
],
"mid": [
"2026593493",
"2132558143",
"1559736362"
],
"abstract": [
"A long-standing puzzle in developmental psychology is how infants imitate gestures they cannot see themselves perform (facial gestures). Two critical issues are: (... |
1901.00811 | 2907904605 | Programming a robot to deal with open-ended tasks remains a challenge, in particular if the robot has to manipulate objects. Launching, grasping, pushing or any other object interaction can be simulated but the corresponding models are not reversible and the robot behavior thus cannot be directly deduced. These behavio... | A proposal to make the sampling process more efficient is to sample in the goal space rather than in the motor space, choosing points in the goal space and using the robot's current inverse model to to reach them, thus generating new samples that allow to further train the inverse model online. This goal babbling'' app... | {
"cite_N": [
"@cite_40",
"@cite_21",
"@cite_25"
],
"mid": [
"2052519881",
"2116086091",
"2101524054"
],
"abstract": [
"We present a neural network approach to early motor learning. The goal is to explore the needs for boot-strapping the control of hand movements in a biologically ... |
1901.00811 | 2907904605 | Programming a robot to deal with open-ended tasks remains a challenge, in particular if the robot has to manipulate objects. Launching, grasping, pushing or any other object interaction can be simulated but the corresponding models are not reversible and the robot behavior thus cannot be directly deduced. These behavio... | In order to simplify inverse model learning, it has been proposed to divide the goal space in several regions according to a spatial segmentation @cite_35 @cite_38 , or in several independant subspaces corresponding to the state of different objects @cite_1 , and learn a different inverse model for each subspace. Anoth... | {
"cite_N": [
"@cite_35",
"@cite_38",
"@cite_28",
"@cite_1",
"@cite_0"
],
"mid": [
"2896027395",
"2004303440",
"2810132790",
"2565678376",
"2963973554"
],
"abstract": [
"Intelligent adaptive curiosity (IAC) was initially introduced as a developmental mechanism allow... |
1901.00811 | 2907904605 | Programming a robot to deal with open-ended tasks remains a challenge, in particular if the robot has to manipulate objects. Launching, grasping, pushing or any other object interaction can be simulated but the corresponding models are not reversible and the robot behavior thus cannot be directly deduced. These behavio... | Unsupervised learning algorithms can extract a small set of primitive actions from direct policy search learning traces @cite_39 , thus allowing to use reinforcement learning with an acquired set of actions. Although this approach works well for problems like navigation where a policy can naturally be described as a se... | {
"cite_N": [
"@cite_39"
],
"mid": [
"2595700172"
],
"abstract": [
"Reinforcement learning (RL) problems are hard to solve in a robotics context as classical algorithms rely on discrete representations of actions and states, but in robotics both are continuous. A discrete set of actions and states... |
1901.00811 | 2907904605 | Programming a robot to deal with open-ended tasks remains a challenge, in particular if the robot has to manipulate objects. Launching, grasping, pushing or any other object interaction can be simulated but the corresponding models are not reversible and the robot behavior thus cannot be directly deduced. These behavio... | Approaches based on divergent evolutionary algorithms such as novelty search @cite_8 , learn a repertoire of actions: they exploit the principles of variation and selection to gradually build a discrete set of actions covering a given - such as a goal space. Since those approaches do not explicitly build an inverse mod... | {
"cite_N": [
"@cite_41",
"@cite_4",
"@cite_20",
"@cite_8"
],
"mid": [
"1974150877",
"2099746672",
"2962687375",
""
],
"abstract": [
"In contrast to the conventional role of evolution in evolutionary computation (EC) as an optimization algorithm, a new class of evolutionary... |
1901.00811 | 2907904605 | Programming a robot to deal with open-ended tasks remains a challenge, in particular if the robot has to manipulate objects. Launching, grasping, pushing or any other object interaction can be simulated but the corresponding models are not reversible and the robot behavior thus cannot be directly deduced. These behavio... | Exploiting this repertoire raises challenges that differ from those of learning an inverse model. Reaching a known goal for which an action is present in the repertoire @cite_8 @cite_41 , is very simple and inexpensive, as the correct action just needs to be selected and executed. Adapting to changes in the problem dom... | {
"cite_N": [
"@cite_7",
"@cite_8",
"@cite_41",
"@cite_32",
"@cite_44",
"@cite_12"
],
"mid": [
"2900424942",
"",
"1974150877",
"758372786",
"2532372495",
"1738827650"
],
"abstract": [
"Learning algorithms are enabling robots to solve increasingly challenging... |
1901.00811 | 2907904605 | Programming a robot to deal with open-ended tasks remains a challenge, in particular if the robot has to manipulate objects. Launching, grasping, pushing or any other object interaction can be simulated but the corresponding models are not reversible and the robot behavior thus cannot be directly deduced. These behavio... | The previously described skill learning techniques are generally performed in a simulator, which acts as a direct model of the robotic system and its environment. Simulation can be much more practical than real robotics, being cheaper, safe from damaging the robot and its environment during experiments, and much faster... | {
"cite_N": [
"@cite_22"
],
"mid": [
"1625577255"
],
"abstract": [
"In robotics, gradient-free optimization algorithms (e.g. evolutionary algorithms) are often used only in simulation because they require the evaluation of many candidate solutions. Nevertheless, solutions obtained in simulation of... |
1901.01007 | 2908087012 | Deep Neural Networks (DNNs) have revolutionized numerous applications, but the demand for ever more performance remains unabated. Scaling DNN computations to larger clusters is generally done by distributing tasks in batch mode using methods such as distributed synchronous SGD. Among the issues with this approach is th... | Much work has addressed the mapping of inference training of CNNs to clusters with programmable accelerators, including @cite_32 @cite_23 . Many frameworks and libraries have been deployed, e.g., MXNet @cite_24 , Caffe @cite_10 , and Tensorflow @cite_2 . These systems hide the complexity of workload decomposition and p... | {
"cite_N": [
"@cite_38",
"@cite_18",
"@cite_22",
"@cite_29",
"@cite_32",
"@cite_24",
"@cite_23",
"@cite_2",
"@cite_10",
"@cite_11"
],
"mid": [
"2557355796",
"",
"2272300165",
"2475840367",
"2769918628",
"2186615578",
"",
"2271840356",
"2... |
1901.01007 | 2908087012 | Deep Neural Networks (DNNs) have revolutionized numerous applications, but the demand for ever more performance remains unabated. Scaling DNN computations to larger clusters is generally done by distributing tasks in batch mode using methods such as distributed synchronous SGD. Among the issues with this approach is th... | Most distributed CNN systems, including TensorFlow and CNTK, are based on the distributed synchronous SGD algorithm (Centralized Parallel SGD algorithm - C-PSGD, see Fig.(A)). The Parameter Server Topology @cite_28 uses a central parameter node connected with multiple worker nodes. Clearly, there are several bottleneck... | {
"cite_N": [
"@cite_28",
"@cite_20",
"@cite_39"
],
"mid": [
"2060393849",
"2963228337",
"2336650964"
],
"abstract": [
"Big data may contain big values, but also brings lots of challenges to the computing theory, architecture, framework, knowledge discovery algorithms, and domain s... |
1901.01007 | 2908087012 | Deep Neural Networks (DNNs) have revolutionized numerous applications, but the demand for ever more performance remains unabated. Scaling DNN computations to larger clusters is generally done by distributing tasks in batch mode using methods such as distributed synchronous SGD. Among the issues with this approach is th... | Fig. shows the design space for mapping CNNs onto distributed nodes. We use terminology introduced by @cite_25 . | {
"cite_N": [
"@cite_25"
],
"mid": [
"2788193959"
],
"abstract": [
"Deep Neural Networks (DNNs) are becoming an important tool in modern computing applications. Accelerating their training is a major challenge and techniques range from distributed algorithms to low-level circuit design. In this su... |
1901.01007 | 2908087012 | Deep Neural Networks (DNNs) have revolutionized numerous applications, but the demand for ever more performance remains unabated. Scaling DNN computations to larger clusters is generally done by distributing tasks in batch mode using methods such as distributed synchronous SGD. Among the issues with this approach is th... | Data parallelism (Fig.(A)) is the most popular approach in CPU and GPU clouds @cite_24 @cite_2 . It is also widely used in existing FPGA clouds, such as Catapult and CDSC @cite_18 . This method has drawbacks as mentioned in Section I. In CNNs, the configurations of each layer, such as kernel size, pooling size, and str... | {
"cite_N": [
"@cite_24",
"@cite_18",
"@cite_2"
],
"mid": [
"2186615578",
"",
"2271840356"
],
"abstract": [
"MXNet is a multi-language machine learning (ML) library to ease the development of ML algorithms, especially for deep neural networks. Embedded in the host language, it blen... |
1901.01007 | 2908087012 | Deep Neural Networks (DNNs) have revolutionized numerous applications, but the demand for ever more performance remains unabated. Scaling DNN computations to larger clusters is generally done by distributing tasks in batch mode using methods such as distributed synchronous SGD. Among the issues with this approach is th... | Layer Parallelism (Fig.(B)) maps layers of the CNN onto individual nodes and pipelines CNN computation. It has been employed by both GPU and FPGA frameworks. In @cite_0 , multiple GPUs are used in a pipelined manner. Each LSTM layer is assigned to a different GPU. After GPU 1 finishes computing layer 1 for the first se... | {
"cite_N": [
"@cite_0",
"@cite_29"
],
"mid": [
"2525778437",
"2475840367"
],
"abstract": [
"Neural Machine Translation (NMT) is an end-to-end learning approach for automated translation, with the potential to overcome many of the weaknesses of conventional phrase-based translation systems... |
1901.00921 | 2907758432 | The solution convergence of Markov Decision Processes (MDPs) can be accelerated by prioritized sweeping of states ranked by their potential impacts to other states. In this paper, we present new heuristics to speed up the solution convergence of MDPs. First, we quantify the level of reachability of every state using th... | Another important heuristic for efficiently solving MDPs is the prioritized sweeping @cite_11 , which has been broadly employed to further speed up the value iteration process. This heuristic evaluates each state and obtains a score based on the state's contribution to the convergence, and then prioritizes sorts all st... | {
"cite_N": [
"@cite_9",
"@cite_4",
"@cite_11"
],
"mid": [
"2159420891",
"2121733891",
"2048226872"
],
"abstract": [
"Prioritized sweeping is a model-based reinforcement learning method that attempts to focus an agent's limited computational resources to achieve a good estimate of ... |
1901.00921 | 2907758432 | The solution convergence of Markov Decision Processes (MDPs) can be accelerated by prioritized sweeping of states ranked by their potential impacts to other states. In this paper, we present new heuristics to speed up the solution convergence of MDPs. First, we quantify the level of reachability of every state using th... | The reachability of state space has been investigated in existing works. For example, the structured reachability analysis @cite_2 of MDPs has been proposed to evaluate whether a state is reachable or not, so that one can restrict the dynamic programming to only reachable states, reducing the computational burden of so... | {
"cite_N": [
"@cite_2"
],
"mid": [
"1545472701"
],
"abstract": [
"Recent research in decision theoretic planning has focussed on making the solution of Markov decision processes (MDPs) more feasible. We develop a family of algorithms for structured reachability analysis of MDPs that are suitable ... |
1901.00921 | 2907758432 | The solution convergence of Markov Decision Processes (MDPs) can be accelerated by prioritized sweeping of states ranked by their potential impacts to other states. In this paper, we present new heuristics to speed up the solution convergence of MDPs. First, we quantify the level of reachability of every state using th... | Important related frameworks for solving MDPs also include compact representations such as linear function representation and approximation @cite_7 @cite_16 used in the policy iteration algorithms. The linear equation based techniques do not exploit regions of uniformity in value functions associated with states, but r... | {
"cite_N": [
"@cite_19",
"@cite_16",
"@cite_7"
],
"mid": [
"1997477668",
"2119567691",
"2028145673"
],
"abstract": [
"Abstract Markov decision processes (MDPs) have proven to be popular models for decision-theoretic planning, but standard dynamic programming algorithms for solving... |
1901.00889 | 2907963126 | Thermal to visible face verification is a challenging problem due to the large domain discrepancy between the modalities. Existing approaches either attempt to synthesize visible faces from thermal faces or extract robust features from these modalities for cross-modal matching. In this paper, we take a different approa... | As described in Figure , traditional thermal to visible face verification methods first extract features from the visible and thermal images and then verify the identity based on the extRacted features. Both hand-crafted and learned features have been investigated in the literature. Hu al @cite_17 proposed a partial le... | {
"cite_N": [
"@cite_30",
"@cite_4",
"@cite_21",
"@cite_17",
"@cite_16",
"@cite_20",
"@cite_11"
],
"mid": [
"2032213187",
"1998165562",
"1991410148",
"2021242760",
"2152788298",
"2204652942",
"2012648224"
],
"abstract": [
"We investigate the performa... |
1901.00889 | 2907963126 | Thermal to visible face verification is a challenging problem due to the large domain discrepancy between the modalities. Existing approaches either attempt to synthesize visible faces from thermal faces or extract robust features from these modalities for cross-modal matching. In this paper, we take a different approa... | Unlike the above mentioned traditional methods, synthesis-based thermal to visible face verification algorithms leverage the synthesized visible faces for verification. Due to the success of CNNs and recently introduced generative adversarial networks (GANs) in synthesizing realistic images, various deep learning-based... | {
"cite_N": [
"@cite_28",
"@cite_9",
"@cite_26",
"@cite_0"
],
"mid": [
"2963639219",
"2566614872",
"2963294002",
"2963276927"
],
"abstract": [
"The large domain discrepancy between faces captured in polarimetric (or conventional) thermal and visible domain makes cross-domai... |
1901.00893 | 2907919191 | We present a method for improving segmentation tasks on images affected by adherent rain drops and streaks. We introduce a novel stereo dataset recorded using a system that allows one lens to be affected by real water droplets while keeping the other lens clear. We train a denoising generator using this dataset and sho... | Generally speaking, the quality of an image can be affected in two ways by bad weather conditions. Firstly, contaminants in the atmosphere, such as falling rain, fog, smog or snow will hinder visibility or partially occlude a scene but do not significantly distort the image. Secondly, adherent contaminants such as wate... | {
"cite_N": [
"@cite_35",
"@cite_36",
"@cite_24",
"@cite_15",
"@cite_34"
],
"mid": [
"2077946335",
"2519481857",
"2509784253",
"1977808497",
"1909316225"
],
"abstract": [
"A novel rain (or snow) streak removal algorithm for stereo video sequences is proposed in this... |
1901.00893 | 2907919191 | We present a method for improving segmentation tasks on images affected by adherent rain drops and streaks. We introduce a novel stereo dataset recorded using a system that allows one lens to be affected by real water droplets while keeping the other lens clear. We train a denoising generator using this dataset and sho... | We base our simple synthetic droplet model on the works of @cite_30 @cite_27 and @cite_20 , by storing proto-droplet normal maps which are subsequently warped and combined at run time using an approach similar to meta-balls @cite_33 . | {
"cite_N": [
"@cite_30",
"@cite_27",
"@cite_33",
"@cite_20"
],
"mid": [
"2038133417",
"1585341626",
"2244686166",
""
],
"abstract": [
"In this paper we present a novel approach to improved image registration in rainy weather situations. To this end, we perform monocular ra... |
1901.00858 | 2907038162 | Breakthroughs in the fields of deep learning and mobile system-on-chips are radically changing the way we use our smartphones. However, deep neural networks inference is still a challenging task for edge AI devices due to the computational overhead on mobile CPUs and a severe drain on the batteries. In this paper, we p... | On the other hand, researchers have also demonstrated many mobile deep learning frameworks, provides various novel features. Deep Compression @cite_16 is a series of techniques aiming to reduce the size of deep neural networks. With pruning, trained quantization and Huffman coding, Deep Compression provides up to @math... | {
"cite_N": [
"@cite_16",
"@cite_11",
"@cite_8"
],
"mid": [
"2119144962",
"2626129225",
"2297325673"
],
"abstract": [
"Neural networks are both computationally intensive and memory intensive, making them difficult to deploy on embedded systems with limited hardware resources. To ad... |
1901.00579 | 2907187671 | As Internet streaming of live content has gained on traditional cable TV viewership, we have also seen significant growth of free live streaming services which illegally provide free access to copyrighted content over the Internet. Some of these services draw millions of viewers each month. Moreover, this viewership ha... | Measuring Online Tracking. @cite_26 presents extensive measurements of online tracking across the Alexa top million websites and presents OpenWPM, the tool we utilize in our work to collect data on illegal stream URLs. Similarly, @cite_14 studies third-party tracking on websites and mobile applications while @cite_16 e... | {
"cite_N": [
"@cite_14",
"@cite_26",
"@cite_16"
],
"mid": [
"2963967433",
"2535603283",
"2272969119"
],
"abstract": [
"Third-party networks collect vast amounts of data about users via websites and mobile applications. Consolidations among tracker companies can significantly incre... |
1901.00579 | 2907187671 | As Internet streaming of live content has gained on traditional cable TV viewership, we have also seen significant growth of free live streaming services which illegally provide free access to copyrighted content over the Internet. Some of these services draw millions of viewers each month. Moreover, this viewership ha... | Illegal Media Streaming. @cite_9 studies security and privacy concerns related to on-demand media streaming services and targets platforms that are known to host illegal content. Specifically, they study over 20 media streaming platforms (e.g., Kodi, Enigma 2, MediaTomb, etc.) and their attack surfaces, and find that t... | {
"cite_N": [
"@cite_8",
"@cite_9",
"@cite_2"
],
"mid": [
"2331802876",
"2792199623",
"2796057488"
],
"abstract": [
"Over recent years, a major shift has occurred in piracy of paid-for content services toward illegal redistribution of live content in real-time over the Internet. Th... |
1901.00579 | 2907187671 | As Internet streaming of live content has gained on traditional cable TV viewership, we have also seen significant growth of free live streaming services which illegally provide free access to copyrighted content over the Internet. Some of these services draw millions of viewers each month. Moreover, this viewership ha... | Illegal Live Media Streaming. @cite_17 studies the ecosystem of free live streaming websites with an analysis of over 5600 live-streaming domains discovered from live-streaming domains through aggregator websites. This study does not focus on user tracking, and instead highlights other aspects of their behavior such as... | {
"cite_N": [
"@cite_17"
],
"mid": [
"2467763829"
],
"abstract": [
"Recent years have seen extensive growth of services enabling free broadcasts of live streams on the Web. Free live streaming (FLIS) services attract millions of viewers and make heavy use of deceptive advertisements. Despite the i... |
1901.00484 | 2908138876 | We describe a novel cross-modal embedding space for actions, named Action2Vec, which combines linguistic cues from class labels with spatio-temporal features derived from video clips. Our approach uses a hierarchical recurrent network to capture the temporal structure of video features. We train our embedding using a j... | One domain where joint models of video and language arise naturally is in the context of video captioning @cite_4 @cite_26 . The task of video captioning faces similar challenges to our problem, but however in captioning the focus is not building and testing a distributed representation but rather focus on the mapping ... | {
"cite_N": [
"@cite_26",
"@cite_4"
],
"mid": [
"1573040851",
"2963843052"
],
"abstract": [
"Automatically describing video content with natural language is a fundamental challenge of computer vision. Re-current Neural Networks (RNNs), which models sequence dynamics, has attracted increasi... |
1901.00484 | 2908138876 | We describe a novel cross-modal embedding space for actions, named Action2Vec, which combines linguistic cues from class labels with spatio-temporal features derived from video clips. Our approach uses a hierarchical recurrent network to capture the temporal structure of video features. We train our embedding using a j... | The final body of related work is recent deep learning approaches that construct video representations for supervised prediction tasks such as action recognition. We build on these approaches in our own work. In particular, our model utilizes the C3D @cite_13 architecture to extract features from video frames. We also ... | {
"cite_N": [
"@cite_36",
"@cite_19",
"@cite_31",
"@cite_13"
],
"mid": [
"2156303437",
"2342662179",
"2619082050",
"1522734439"
],
"abstract": [
"We investigate architectures of discriminatively trained deep Convolutional Networks (ConvNets) for action recognition in video.... |
1901.00568 | 2907940234 | Three Dimensional Integrated Circuits (3D IC) offer lower power consumption, higher performance, higher bandwidth, and scalability over the conventional two dimensional ICs. Through-Silicon Via (TSV) is one of the fabrication mechanisms that connects stacked dies to each other. The large size of TSVs and the proximity ... | In the context of (2D NoC), there are plenty of works that target power consumption @cite_5 @cite_2 @cite_25 , reliability @cite_5 , security @cite_17 , or performance @cite_5 of the interconnections. Particularly, crosstalk minimization methods can be classified in three categories: physical level, transistor level an... | {
"cite_N": [
"@cite_11",
"@cite_4",
"@cite_26",
"@cite_7",
"@cite_9",
"@cite_0",
"@cite_2",
"@cite_5",
"@cite_15",
"@cite_25",
"@cite_17"
],
"mid": [
"2130315896",
"2153878017",
"2115468262",
"",
"2117549011",
"1999402023",
"2528200010",
... |
1901.00568 | 2907940234 | Three Dimensional Integrated Circuits (3D IC) offer lower power consumption, higher performance, higher bandwidth, and scalability over the conventional two dimensional ICs. Through-Silicon Via (TSV) is one of the fabrication mechanisms that connects stacked dies to each other. The large size of TSVs and the proximity ... | Although the above approaches may cope with crosstalk in 2D ICs, they cannot be directly applied in 3D technologies because the additional dimension makes consequential differences in crosstalk problem analysis. Gathering the long and thick TSVs causes new reliability issues which have been studied recently @cite_24 @c... | {
"cite_N": [
"@cite_22",
"@cite_8",
"@cite_1",
"@cite_3",
"@cite_6",
"@cite_24",
"@cite_13",
"@cite_12"
],
"mid": [
"",
"1999816809",
"2155351061",
"2009570976",
"2003631910",
"2004420027",
"2069756028",
"2017619982"
],
"abstract": [
"",
... |
1901.00568 | 2907940234 | Three Dimensional Integrated Circuits (3D IC) offer lower power consumption, higher performance, higher bandwidth, and scalability over the conventional two dimensional ICs. Through-Silicon Via (TSV) is one of the fabrication mechanisms that connects stacked dies to each other. The large size of TSVs and the proximity ... | Increasing TSV distances from each other, shielding TSVs, inserting buffers at the victim side, inserting buffers, decreasing driver size at the aggressor side, and increasing load at the wires are the mechanisms examined in @cite_20 to mitigate TSVs crosstalk noise. According to their experiments, unlike 2D wires, inc... | {
"cite_N": [
"@cite_20"
],
"mid": [
"2133279383"
],
"abstract": [
"This paper studies TSV-to-TSV coupling in 3D ICs. A full-chip SI analysis flow is proposed based on the proposed coupling model. Analysis results show that TSVs cause significant coupling noise and timing problems despite that TSV... |
1901.00568 | 2907940234 | Three Dimensional Integrated Circuits (3D IC) offer lower power consumption, higher performance, higher bandwidth, and scalability over the conventional two dimensional ICs. Through-Silicon Via (TSV) is one of the fabrication mechanisms that connects stacked dies to each other. The large size of TSVs and the proximity ... | RTL mechanisms in 3D IC have been proposed and experimented recently. @cite_12 proposed a coding scheme that reduces the maximum crosstalk about 28 The authors in @cite_8 introduce use of less adjacent transition code along with transition signaling to minimize the number of transitions. Furthermore, 3DLAT reduces high... | {
"cite_N": [
"@cite_12",
"@cite_8"
],
"mid": [
"2017619982",
"1999816809"
],
"abstract": [
"In 3D VLSI, through-silicon vias (TSVs) are relatively large, and closely spaced. This results in a situation in which noise on one or more TSVs may deteriorate the delay and signal integrity of ne... |
1901.00512 | 2908315377 | A brain-computer interface (BCI) based on the motor imagery (MI) paradigm translates one's motor intention into a control signal by classifying the Electroencephalogram (EEG) signal of different tasks. However, most existing systems either (i) use a high-quality algorithm to train the data off-line and run only classif... | Improved techniques for using coresets for distributed data and low communication on the cloud, with both theoretical guarantees and experimental results were recently suggested in conferences such as @cite_53 @cite_22 . Classical techniques such as Frank-Wolfe @cite_44 and semi-definite programming @cite_45 appear to ... | {
"cite_N": [
"@cite_22",
"@cite_53",
"@cite_29",
"@cite_32",
"@cite_44",
"@cite_19",
"@cite_24",
"@cite_45",
"@cite_59",
"@cite_46",
"@cite_25",
"@cite_11"
],
"mid": [
"2088424151",
"2003895866",
"1813460488",
"2157754768",
"2109706083",
"20... |
1901.00366 | 2908467437 | Knowledge Distillation (KD) has been used in image classification for model compression. However, rare studies apply this technology on single-stage object detectors. Focal loss shows that the accumulated errors of easily-classified samples dominate the overall loss in the training process. This problem is also encount... | Many works are proposed to accelerate the convolution neural network due to the demand for practice applications. Knowledge transferring is one approach that transfers knowledge from the teacher model to the student model. Previous work explores this area by representing knowledge in different forms. FitNet @cite_3 mak... | {
"cite_N": [
"@cite_3",
"@cite_2"
],
"mid": [
"1690739335",
"1821462560"
],
"abstract": [
"While depth tends to improve network performances, it also makes gradient-based training more difficult since deeper networks tend to be more non-linear. The recently proposed knowledge distillation... |
1901.00366 | 2908467437 | Knowledge Distillation (KD) has been used in image classification for model compression. However, rare studies apply this technology on single-stage object detectors. Focal loss shows that the accumulated errors of easily-classified samples dominate the overall loss in the training process. This problem is also encount... | Recently, model compression has been studied to facilitate the application of cnn-based object detector in devices with limited computation resources. Chen al @cite_17 utilizes soft targets to guide the student model in both region proposal network and region convolution neural network and balance the positive and nega... | {
"cite_N": [
"@cite_7",
"@cite_17"
],
"mid": [
"2750432752",
"2750784772"
],
"abstract": [
"Current CNN based object detectors need initialization from pre-trained ImageNet classification models, which are usually time-consuming. In this paper, we present a fully convolutional feature mim... |
1907.04251 | 2959979976 | We propose an algorithm for low rank matrix completion for matrices with binary entries which obtains explicit binary factors. Our algorithm, which we call TBMC (), gives interpretable output in the form of binary factors which represent a decomposition of the matrix into tiles. Our approach is inspired by a popular al... | We propose TBMC (Tiling for Binary Matrix Completion), a low rank binary matrix completion algorithm ( sec:TBMC ). The algorithm is inspired by the approach in @cite_9 for BMF by recursively partitioning the database by means of rank-one approximations. In particular, we propose using an LP rank-one approximation for m... | {
"cite_N": [
"@cite_9",
"@cite_14",
"@cite_25"
],
"mid": [
"2167851099",
"1977877565",
"1995565521"
],
"abstract": [
"This article presents the design and implementation of a software tool, PROXIMUS, for error-bounded approximation of high-dimensional binary attributed datasets ba... |
1907.03928 | 2958621236 | Probabilistic game structures combine both nondeterminism and stochasticity, where players repeatedly take actions simultaneously to move to the next state of the concurrent game. Probabilistic alternating simulation is an important tool to compare the behaviour of different probabilistic game structures. In this paper... | Segala and Lynch @cite_5 introduce a probabilistic simulation relation which preserves probabilistic computation tree logic (PCTL) formulas without negation and existential quantification. Segala introduces the notion of probabilistic forward simulation, which relates states to probability distributions over states and... | {
"cite_N": [
"@cite_4",
"@cite_7",
"@cite_6",
"@cite_24",
"@cite_27",
"@cite_5",
"@cite_10"
],
"mid": [
"1505927408",
"2037507558",
"",
"2152206644",
"2168098347",
"138290785",
"1548037201"
],
"abstract": [
"We give logical characterizations of bisi... |
1907.03928 | 2958621236 | Probabilistic game structures combine both nondeterminism and stochasticity, where players repeatedly take actions simultaneously to move to the next state of the concurrent game. Probabilistic alternating simulation is an important tool to compare the behaviour of different probabilistic game structures. In this paper... | Metric-based simulation on game structures have been studied by de @cite_15 regarding the probability of winning games whose goals are expressed in quantitative @math -calculus (qMu) @cite_1 . Two states are equivalent if the players can win the same games with the same probability from both states, and among states ca... | {
"cite_N": [
"@cite_13",
"@cite_15",
"@cite_1"
],
"mid": [
"197133740",
"2038193812",
"2091027967"
],
"abstract": [
"A method of producing long life precision abrasive articles for use in met al removal operations. In this method abrasive particles are impinged against the inner s... |
1907.03928 | 2958621236 | Probabilistic game structures combine both nondeterminism and stochasticity, where players repeatedly take actions simultaneously to move to the next state of the concurrent game. Probabilistic alternating simulation is an important tool to compare the behaviour of different probabilistic game structures. In this paper... | More recently, algorithmic verification of turn-based and concurrent games have been implemented as an extension of PRISM @cite_23 @cite_21 . The properties can be specified as state formulas, path formulas and reward formulas. The verification procedure requires solving matrix games for concurrent game structures, and... | {
"cite_N": [
"@cite_15",
"@cite_21",
"@cite_23",
"@cite_2"
],
"mid": [
"2038193812",
"2886584225",
"2771041648",
"2060155289"
],
"abstract": [
"We consider two-player games played over finite state spaces for an infinite number of rounds. At each state, the players simulta... |
1907.03993 | 2961402400 | Many complex networks in the real world have community structures -- groups of well-connected nodes with important functional roles. It has been well recognized that the identification of communities bears numerous practical applications. While existing approaches mainly apply statistical or graph theoretical combinato... | Ricci curvature on general spaces without Riemannian structures has been recently studied, in the work of Ollivier @cite_67 @cite_12 on Markov chains, and Bakry and Emery @cite_13 , Lott, Villani @cite_24 , Bonciocat and Sturm @cite_62 @cite_3 on general metric spaces. Ricci curvature based on optimal transportation th... | {
"cite_N": [
"@cite_36",
"@cite_54",
"@cite_3",
"@cite_5",
"@cite_15",
"@cite_67",
"@cite_4",
"@cite_60",
"@cite_46",
"@cite_37",
"@cite_6",
"@cite_19",
"@cite_12",
"@cite_13",
"@cite_62",
"@cite_14",
"@cite_24",
"@cite_59",
"@cite_51"
],
... |
1907.03993 | 2961402400 | Many complex networks in the real world have community structures -- groups of well-connected nodes with important functional roles. It has been well recognized that the identification of communities bears numerous practical applications. While existing approaches mainly apply statistical or graph theoretical combinato... | Unlike discrete Ricci curvature, discrete Ricci flow has not been studied as much. Chow and Luo introduced the first discrete Ricci flow on surfaces @cite_65 . In @cite_32 , Weber al suggested applying Forman-Ricci flow for anomaly detection in the complex network. In @cite_38 , Ni al used the Ollivier-Ricci curvature ... | {
"cite_N": [
"@cite_38",
"@cite_65",
"@cite_32"
],
"mid": [
"2891096760",
"2143075606",
"2963757685"
],
"abstract": [
"In this paper, we consider the problem of approximately aligning matching two graphs. Given two graphs (G_ 1 =(V_ 1 ,E_ 1 ) ) and (G_ 2 =(V_ 2 ,E_ 2 ) ), the obje... |
1907.03993 | 2961402400 | Many complex networks in the real world have community structures -- groups of well-connected nodes with important functional roles. It has been well recognized that the identification of communities bears numerous practical applications. While existing approaches mainly apply statistical or graph theoretical combinato... | Taking a geometric view of complex networks is an emerging trend, as shown in a number of recent work. For example, the community structures were used as a coarse version of its embedding in a hidden space with hyperbolic geometry @cite_29 . Topological data analysis, a typical geometric approach for data analysis, has... | {
"cite_N": [
"@cite_29",
"@cite_8"
],
"mid": [
"2888988444",
"2896791121"
],
"abstract": [
"We show that the community structure of a network can be used as a coarse version of its embedding in a hidden space with hyperbolic geometry. The finding emerges from a systematic analysis of seve... |
1907.03956 | 2958276832 | This paper presents planning algorithms for a robotic manipulator with a fixed base in order to grasp a target object in cluttered environments. We consider a configuration of objects in a confined space with a high density so no collision-free path to the target exists. The robot must relocate some objects to retrieve... | The work presented in @cite_13 proposes a planning framework to grasp a target in cluttered and known environments. It removes obstacles that are in the shortest path of the end-effector to the target (like Fig. -L). Although this method finds the distance-optimal path, some obstacles could have to be removed unnecessa... | {
"cite_N": [
"@cite_20",
"@cite_13",
"@cite_12",
"@cite_17"
],
"mid": [
"2963566599",
"1989021449",
"2141841102",
"1608892862"
],
"abstract": [
"Planning motions to grasp an object in cluttered and uncertain environments is a challenging task, particularly when a collision... |
1907.03956 | 2958276832 | This paper presents planning algorithms for a robotic manipulator with a fixed base in order to grasp a target object in cluttered environments. We consider a configuration of objects in a confined space with a high density so no collision-free path to the target exists. The robot must relocate some objects to retrieve... | Some recent work considers partially known environments. The algorithm proposed in @cite_7 computes a sequence of objects to be removed while minimizing the expected time to find a hidden target. The strength of this work is the mathematical formalization of the search and grasp planning problem. However, the algorithm... | {
"cite_N": [
"@cite_9",
"@cite_18",
"@cite_1",
"@cite_7"
],
"mid": [
"2561079189",
"1600604169",
"2419360507",
"2084006613"
],
"abstract": [
"We study the problem of objects search in clutter. In cluttered environments, partial occlusion among objects prevents vision syste... |
1907.03956 | 2958276832 | This paper presents planning algorithms for a robotic manipulator with a fixed base in order to grasp a target object in cluttered environments. We consider a configuration of objects in a confined space with a high density so no collision-free path to the target exists. The robot must relocate some objects to retrieve... | Among these, no work has formulated the problem as an optimization problem whose objective value is the number of obstacles to be relocated. The methods presented in these work require substantial planning time in clutter. The examples that we will consider are significantly more cluttered so we need faster planning al... | {
"cite_N": [
"@cite_10",
"@cite_8"
],
"mid": [
"2114476723",
"2968156685"
],
"abstract": [
"This paper presents further improvements on the earlier vector field histogram (VFH) method developed by Borenstein-Koren (1991) for real-time mobile robot obstacle avoidance. The enhanced method, ... |
1907.04068 | 2961307394 | We consider the hypothesis testing problem of detecting conditional dependence, with a focus on high-dimensional feature spaces. Our contribution is a new test statistic based on samples from a generative adversarial network designed to approximate directly a conditional distribution that encodes the null hypothesis, i... | A recent favoured line of research has characterized conditional independence in a (RKHS) @cite_26 @cite_24 . The dependence between variables is assessed considering all moments of the joint distributions which potentially captures finer differences between them. @cite_26 uses a measure of partial association in a RKH... | {
"cite_N": [
"@cite_24",
"@cite_5",
"@cite_29",
"@cite_26"
],
"mid": [
"609741286",
"2601307951",
"2964340499",
"2951039901"
],
"abstract": [
"Determining conditional independence (CI) relationships between random variables is a challenging but important task for problems ... |
1907.04214 | 2956996691 | An optimal feedback controller for a given Markov decision process (MDP) can in principle be synthesized by value or policy iteration. However, if the system dynamics and the reward function are unknown, a learning agent must discover an optimal controller via direct interaction with the environment. Such interactive d... | Apart from computational advantages, information-theoretic approaches provide a solid framework for describing and studying aspects of intelligent behavior @cite_27 , from autonomy @cite_39 and curiosity @cite_42 to bounded rationality @cite_17 and game theory @cite_11 . | {
"cite_N": [
"@cite_11",
"@cite_42",
"@cite_39",
"@cite_27",
"@cite_17"
],
"mid": [
"1487708124",
"2020920737",
"2054162326",
"2239029832",
"2211766770"
],
"abstract": [
"A long-running difficulty with conventional game theory has been how to modify it to accommoda... |
1907.04214 | 2956996691 | An optimal feedback controller for a given Markov decision process (MDP) can in principle be synthesized by value or policy iteration. However, if the system dynamics and the reward function are unknown, a learning agent must discover an optimal controller via direct interaction with the environment. Such interactive d... | Entropic proximal mappings were introduced in @cite_29 as a general framework for constructing approximation and smoothing schemes for optimization problem. Problem formulation presented here can be considered as an application of this general theory to policy optimization in Markov decision processes. Following the re... | {
"cite_N": [
"@cite_14",
"@cite_22",
"@cite_29",
"@cite_44",
"@cite_34",
"@cite_12"
],
"mid": [
"1771410628",
"2619268125",
"2009274429",
"1499669280",
"2016384870",
"1505731132"
],
"abstract": [
"In this article, we describe a method for optimizing control... |
1907.04214 | 2956996691 | An optimal feedback controller for a given Markov decision process (MDP) can in principle be synthesized by value or policy iteration. However, if the system dynamics and the reward function are unknown, a learning agent must discover an optimal controller via direct interaction with the environment. Such interactive d... | An alternative proximal reinforcement learning scheme was introduced in @cite_28 based on the extragradient method for solving variational inequalities and leveraging operator splitting techniques. Although the idea of exploiting proximal maps and updates in the primal and dual spaces is similar to ours, regularization... | {
"cite_N": [
"@cite_28"
],
"mid": [
"1835716857"
],
"abstract": [
"In this paper, we set forth a new vision of reinforcement learning developed by us over the past few years, one that yields mathematically rigorous solutions to longstanding important questions that have remained unresolved: (i) h... |
1907.04072 | 2957984643 | Online social media platforms have made the world more connected than ever before, thereby making it easier for everyone to spread their content across a wide variety of audiences. Twitter is one such popular platform where people publish tweets to spread their messages to everyone. Twitter allows users to Retweet othe... | : The problem of fake and spam tweets is not new. Many solutions have been proposed to tackle this problem. @cite_16 showed that the network structure of spammers and non-spammers is different, and also tracked the life cycle of endogenous Twitter content. @cite_15 conducted a comprehensive evaluation of several machin... | {
"cite_N": [
"@cite_14",
"@cite_15",
"@cite_16",
"@cite_12",
"@cite_11"
],
"mid": [
"2280128323",
"1526831942",
"2074835059",
"1796766288",
"1085730058"
],
"abstract": [
"Viral marketing, marketing techniques that use pre-existing social networks, has experienced a... |
1907.04072 | 2957984643 | Online social media platforms have made the world more connected than ever before, thereby making it easier for everyone to spread their content across a wide variety of audiences. Twitter is one such popular platform where people publish tweets to spread their messages to everyone. Twitter allows users to Retweet othe... | : Blackmarket services have recently received considerable attention due to the increase in the number of users using them. Analysis of such underground services was first documented in @cite_2 where the authors examined the properties of social networks formed for blackmarket services. @cite_4 proposed DetectVC which ... | {
"cite_N": [
"@cite_0",
"@cite_10",
"@cite_4",
"@cite_2"
],
"mid": [
"2907374582",
"2809692089",
"2572911049",
"2122551442"
],
"abstract": [
"Twitter's popularity has fostered the emergence of various illegal user activities - one such activity is to artificially bolster v... |
1907.04072 | 2957984643 | Online social media platforms have made the world more connected than ever before, thereby making it easier for everyone to spread their content across a wide variety of audiences. Twitter is one such popular platform where people publish tweets to spread their messages to everyone. Twitter allows users to Retweet othe... | : Multitask learning is used whenever we have two or more similar tasks to optimise together. Most of the related studies on multitask learning are based on how the tasks can be better learned together. @cite_7 classified multitask learning models into five types and reported the characteristics of each approach. Cross... | {
"cite_N": [
"@cite_13",
"@cite_7",
"@cite_8"
],
"mid": [
"2903014193",
"2742079690",
"2963877604"
],
"abstract": [
"In several natural language tasks, labeled sequences are available in separate domains (say, languages), but the goal is to label sequences with mixed domain (such ... |
1901.00326 | 2906748410 | In this paper, we propose a novel method to incorporate partial evidence in the inference of deep convolutional neural networks. Contrary to the existing, top performing methods, which either iteratively modify the input of the network or exploit external label taxonomy to take the partial evidence into account, we add... | : Exploiting additional contextual cues in visual recognition tasks gained a lot of attention from the computer vision community @cite_14 @cite_24 @cite_20 . Contextual information related to semantics was used to improve object detection @cite_23 . Social media meta-data was also used in the contest of multilabel imag... | {
"cite_N": [
"@cite_24",
"@cite_14",
"@cite_23",
"@cite_20"
],
"mid": [
"2160254296",
"2141364309",
"2125215748",
"1908139891"
],
"abstract": [
"In this work we introduce a novel approach to object categorization that incorporates two types of context-co-occurrence and rel... |
1901.00326 | 2906748410 | In this paper, we propose a novel method to incorporate partial evidence in the inference of deep convolutional neural networks. Contrary to the existing, top performing methods, which either iteratively modify the input of the network or exploit external label taxonomy to take the partial evidence into account, we add... | : Some authors proposed to model co-occurrence of labels available at training time to improve recognition performance @cite_22 . @cite_26 , on the other hand, uses a special structure to store the relations between the labels using a graph designed specifically to capture semantic similarities between the labels. Othe... | {
"cite_N": [
"@cite_26",
"@cite_11",
"@cite_22",
"@cite_0",
"@cite_13",
"@cite_20",
"@cite_17"
],
"mid": [
"64813323",
"2125204570",
"2706729717",
"2135166986",
"2153419563",
"1908139891",
"2098728436"
],
"abstract": [
"In this paper we study how to... |
1901.00326 | 2906748410 | In this paper, we propose a novel method to incorporate partial evidence in the inference of deep convolutional neural networks. Contrary to the existing, top performing methods, which either iteratively modify the input of the network or exploit external label taxonomy to take the partial evidence into account, we add... | : Somehow related to our work is a recently thriving area of multi-task learning. Motivated by the phenomenon of catastrophic forgetting, multi-task learning tries to address the problem of lifelong learning and adaptation of a neural network to a set of changing tasks while preserving network's structure. In @cite_6 ,... | {
"cite_N": [
"@cite_19",
"@cite_6"
],
"mid": [
"2963211188",
"2605043629"
],
"abstract": [
"There is a growing interest in learning data representations that work well for many different types of problems and data. In this paper, we look in particular at the task of learning a single visu... |
1901.00326 | 2906748410 | In this paper, we propose a novel method to incorporate partial evidence in the inference of deep convolutional neural networks. Contrary to the existing, top performing methods, which either iteratively modify the input of the network or exploit external label taxonomy to take the partial evidence into account, we add... | Finally, the most relevant to the work presented in this paper are two methods proposed by Hu al @cite_15 and Wang al @cite_7 . Both of them address the problem of visual tasks in the presence of partial evidence. | {
"cite_N": [
"@cite_15",
"@cite_7"
],
"mid": [
"2963513598",
"2963175631"
],
"abstract": [
"Images of scenes have various objects as well as abundant attributes, and diverse levels of visual categorization are possible. A natural image could be assigned with finegrained labels that descri... |
1901.00326 | 2906748410 | In this paper, we propose a novel method to incorporate partial evidence in the inference of deep convolutional neural networks. Contrary to the existing, top performing methods, which either iteratively modify the input of the network or exploit external label taxonomy to take the partial evidence into account, we add... | Hu al @cite_15 tackles this challenge by proposing a Structured Inference Neural Network (SINN). The SINN method is designed to discover the hierarchical structure of labels but can be used in partial evidence setup if labels at given hierarchy are clamped at inference. The SINN model, however, uses CNN and LSTM to dis... | {
"cite_N": [
"@cite_15"
],
"mid": [
"2963513598"
],
"abstract": [
"Images of scenes have various objects as well as abundant attributes, and diverse levels of visual categorization are possible. A natural image could be assigned with finegrained labels that describe major components, coarsegraine... |
1901.00326 | 2906748410 | In this paper, we propose a novel method to incorporate partial evidence in the inference of deep convolutional neural networks. Contrary to the existing, top performing methods, which either iteratively modify the input of the network or exploit external label taxonomy to take the partial evidence into account, we add... | The FeedbackProp proposed by Wang al @cite_7 , on the other hand, uses an iterative procedure which is applied at inference time. The idea is to modify network activations to maximize probabilities of labels under the partial evidence. The method does not require to re-train base model. However, due to iterative proced... | {
"cite_N": [
"@cite_7"
],
"mid": [
"2963175631"
],
"abstract": [
"We propose an inference procedure for deep convolutional neural networks (CNNs) when partial evidence is available. Our method consists of a general feedback-based propagation approach (feedback-prop) that boosts the prediction acc... |
1901.00117 | 2907704766 | Robust Policy Search is the problem of learning policies that do not degrade in performance when subject to unseen environment model parameters. It is particularly relevant for transferring policies learned in a simulation environment to the real world. Several existing approaches involve sampling large batches of traj... | @cite_29 learn controllers with a specific funtional form using trajectories sampled for parameters drawn from an ensemble, and optimize for the average case performance. @cite_28 propose EPOpt, which learns a Neural Network (NN) policy using a model-free DRL algorithm, but on simulated domains sampled from an ensemble... | {
"cite_N": [
"@cite_28",
"@cite_29",
"@cite_1",
"@cite_17"
],
"mid": [
"2964173023",
"1966784014",
"2605102758",
"2963614114"
],
"abstract": [
"Sample complexity and safety are major challenges when learning policies with reinforcement learning for real-world tasks, especi... |
1901.00117 | 2907704766 | Robust Policy Search is the problem of learning policies that do not degrade in performance when subject to unseen environment model parameters. It is particularly relevant for transferring policies learned in a simulation environment to the real world. Several existing approaches involve sampling large batches of traj... | A recent work that learns from an ensemble of models is ( @cite_13 ), but the ensemble here consists of learned DNN models of the dynamics for use in Model Based RL, rather than being induced by changing physical properties of the environment. A similar ensemble generated by perturbing an already learned model is used ... | {
"cite_N": [
"@cite_13",
"@cite_3"
],
"mid": [
"2963846183",
"2205975260"
],
"abstract": [
"Model-free reinforcement learning (RL) methods are succeeding in a growing number of tasks, aided by recent advances in deep learning. They tend to suffer from high sample complexity, however, whic... |
1901.00117 | 2907704766 | Robust Policy Search is the problem of learning policies that do not degrade in performance when subject to unseen environment model parameters. It is particularly relevant for transferring policies learned in a simulation environment to the real world. Several existing approaches involve sampling large batches of traj... | Although EPOpt uses only an appropriate subset of models to train on, none of the above approaches consider ways to sample trajectories only as necessary. Our proposed framework employs active learning to decide with data from only a few model parameters the models for which the agent requires more training. Active sam... | {
"cite_N": [
"@cite_27"
],
"mid": [
"2963488722"
],
"abstract": [
"One of the long-standing challenges in Artificial Intelligence for learning goal-directed behavior is to build a single agent which can solve multiple tasks. Recent progress in multi-task learning for goal-directed sequential prob... |
1901.00148 | 2907715846 | Existing pose estimation approaches fall into two categories: single-stage and multi-stage methods. While multi-stage methods are seemingly more suited for the task, their performance in current practice is not as good as single-stage methods. This work studies this issue. We argue that the current multi-stage methods'... | Single-Stage Approach Single-stage methods @cite_34 @cite_1 @cite_11 @cite_27 are based on backbone networks that are well tuned on image classification tasks, such as VGG @cite_19 or ResNet @cite_10 . Papandreou al @cite_34 designs a network to generate heat maps as well as their relative offsets to get the final pred... | {
"cite_N": [
"@cite_1",
"@cite_19",
"@cite_27",
"@cite_34",
"@cite_10",
"@cite_11"
],
"mid": [
"",
"1686810756",
"2796779902",
"",
"2194775991",
"2769331938"
],
"abstract": [
"",
"In this work we investigate the effect of the convolutional network depth... |
1901.00148 | 2907715846 | Existing pose estimation approaches fall into two categories: single-stage and multi-stage methods. While multi-stage methods are seemingly more suited for the task, their performance in current practice is not as good as single-stage methods. This work studies this issue. We argue that the current multi-stage methods'... | Multi-Stage Approach Multi-Stage methods @cite_39 @cite_24 @cite_3 @cite_36 @cite_16 @cite_32 aim to produce increasingly refined estimation. They can be bottom-up or top-down. In contrary, single-stage methods are all top-down. | {
"cite_N": [
"@cite_36",
"@cite_32",
"@cite_3",
"@cite_39",
"@cite_24",
"@cite_16"
],
"mid": [
"2307770531",
"2795262365",
"2555751471",
"2964304707",
"2559085405",
"2742737904"
],
"abstract": [
"This work introduces a novel convolutional network architectu... |
1901.00148 | 2907715846 | Existing pose estimation approaches fall into two categories: single-stage and multi-stage methods. While multi-stage methods are seemingly more suited for the task, their performance in current practice is not as good as single-stage methods. This work studies this issue. We argue that the current multi-stage methods'... | Bottom-up methods firstly predict individual joints in the image and then associate these joints into human instances. Cao al @cite_24 employs a VGG-19 @cite_19 network as a feature encoder, then the output features go through a multi-stage network resulting in heat maps and associations of the keypoints. Newell al @ci... | {
"cite_N": [
"@cite_24",
"@cite_19",
"@cite_3"
],
"mid": [
"2559085405",
"1686810756",
"2555751471"
],
"abstract": [
"We present an approach to efficiently detect the 2D pose of multiple people in an image. The approach uses a nonparametric representation, which we refer to as Par... |
1901.00148 | 2907715846 | Existing pose estimation approaches fall into two categories: single-stage and multi-stage methods. While multi-stage methods are seemingly more suited for the task, their performance in current practice is not as good as single-stage methods. This work studies this issue. We argue that the current multi-stage methods'... | Top-down approaches first locate the persons using detectors @cite_5 @cite_26 @cite_37 . And a single person pose estimator is then used to predict the keypoints locations. Wei al @cite_39 employs deep convolutional neural networks as feature encoder to estimate human pose. This work designs a sequential architecture c... | {
"cite_N": [
"@cite_38",
"@cite_37",
"@cite_26",
"@cite_4",
"@cite_36",
"@cite_28",
"@cite_32",
"@cite_34",
"@cite_39",
"@cite_27",
"@cite_5",
"@cite_31",
"@cite_16",
"@cite_20",
"@cite_11"
],
"mid": [
"",
"2797527871",
"2565639579",
"20... |
1901.00282 | 2907361665 | In the presence of large sets of labeled data, Deep Learning (DL) has accomplished extraordinary triumphs in the avenue of computer vision, particularly in object classification and recognition tasks. However, DL cannot always perform well when the training and testing images come from different distributions or in the... | There have been many domain adaptation methods @cite_28 @cite_17 @cite_25 @cite_37 @cite_0 @cite_10 @cite_1 @cite_31 proposed in recent years to solve the problem of domain bias. All the methods can be categorized into two main categories, Conventional Domain Adaptation and Deep Domain Adaptation methods. The conventio... | {
"cite_N": [
"@cite_37",
"@cite_28",
"@cite_1",
"@cite_0",
"@cite_31",
"@cite_10",
"@cite_25",
"@cite_17"
],
"mid": [
"2963403405",
"2963767194",
"2214409633",
"2963864946",
"2964288524",
"2962687275",
"2964057616",
"2963993484"
],
"abstract": [... |
1901.00282 | 2907361665 | In the presence of large sets of labeled data, Deep Learning (DL) has accomplished extraordinary triumphs in the avenue of computer vision, particularly in object classification and recognition tasks. However, DL cannot always perform well when the training and testing images come from different distributions or in the... | Obtaining the features using deep neural network even without adaptation technique outperform the conventional DA methods by large margin. However, the results achieved with the Deep Convolutional Activation Features (DeCAF) @cite_20 even without using any adaptation technique to the target data are remarkably better t... | {
"cite_N": [
"@cite_20"
],
"mid": [
"2155541015"
],
"abstract": [
"We evaluate whether features extracted from the activation of a deep convolutional network trained in a fully supervised fashion on a large, fixed set of object recognition tasks can be repurposed to novel generic tasks. Our gener... |
1901.00282 | 2907361665 | In the presence of large sets of labeled data, Deep Learning (DL) has accomplished extraordinary triumphs in the avenue of computer vision, particularly in object classification and recognition tasks. However, DL cannot always perform well when the training and testing images come from different distributions or in the... | MMD is a popular metric for measuring the distributions of source and target samples. @cite_32 proposed the Deep Domain Confusion (DDC) domain adaptation framework based on a confusion layer for the discrepancy between source and target data. In @cite_1 , the previous work is extended by introducing soft label distribu... | {
"cite_N": [
"@cite_30",
"@cite_4",
"@cite_1",
"@cite_32",
"@cite_16",
"@cite_34"
],
"mid": [
"2627183927",
"2159291411",
"2214409633",
"1565327149",
"2279034837",
"2964278684"
],
"abstract": [
"In recent years, deep neural networks have emerged as a domina... |
1901.00140 | 2907429861 | Low-rank matrix factorization (LRMF) has received much popularity owing to its successful applications in both computer vision and data mining. By assuming the noise term to come from a Gaussian, Laplace or a mixture of Gaussian distributions, significant efforts have been made on optimizing the (weighted) @math or @ma... | Recently, the research community began to focus on probabilistic extensions of robust matrix factorizations. Generally speaking, it is assumed that @math , where @math is a noise matrix. Lakshminarayanan @cite_9 replaced Gaussian noise with Gaussian scale mixture noise. Nevertheless, it may be ineffective when processi... | {
"cite_N": [
"@cite_9",
"@cite_1",
"@cite_12",
"@cite_4"
],
"mid": [
"2188214461",
"199797433",
"2492899067",
"2138507544"
],
"abstract": [
"We analyse the noise arising in collaborative filtering when formalised as a probabilistic matrix factorisation problem. We show emp... |
1901.00140 | 2907429861 | Low-rank matrix factorization (LRMF) has received much popularity owing to its successful applications in both computer vision and data mining. By assuming the noise term to come from a Gaussian, Laplace or a mixture of Gaussian distributions, significant efforts have been made on optimizing the (weighted) @math or @ma... | On the other hand, robust principle component analysis (robust PCA) @cite_2 considers an issue that is similar to LRMF, that is, The underlying assumption of robust PCA is that the original data can be decomposed into the sum of a low-rank matrix and a sparse outlier matrix (i.e., the number of non-zero elements in @ma... | {
"cite_N": [
"@cite_10",
"@cite_2"
],
"mid": [
"2045983409",
"2145962650"
],
"abstract": [
"In this paper, we study the problem of recovering a low-rank matrix (the principal components) from a high-dimensional data matrix despite both small entry-wise noise and gross sparse errors. Recen... |
1901.00132 | 2963652257 | In 5G research, it is traditionally assumed that vertical industries (a.k.a verticals) set the performance requirements for the services they want to offer to mobile users, and the mobile operators alone are in charge of orchestrating their resources so as to meet such requirements. Motivated by the observation that su... | Unlike their fourth-generation counterparts, 5G networks will not only transport data, but also process them. Network, computing, and memory resources controlled by mobile network operators (MNOs), will concurrently support multiple services, under the network slicing paradigm @cite_0 @cite_6 . It is universally expect... | {
"cite_N": [
"@cite_3",
"@cite_6",
"@cite_0",
"@cite_10",
"@cite_11"
],
"mid": [
"2612074600",
"2605961225",
"2604174486",
"2744111766",
"2805731797"
],
"abstract": [
"5G is envisioned to be a multi-service network supporting a wide range of verticals with a divers... |
1901.00132 | 2963652257 | In 5G research, it is traditionally assumed that vertical industries (a.k.a verticals) set the performance requirements for the services they want to offer to mobile users, and the mobile operators alone are in charge of orchestrating their resources so as to meet such requirements. Motivated by the observation that su... | Our purpose in this paper is to study a different model of interaction between vertical industries (henceforth verticals) and MNOs, whereby verticals provide not only the target KPIs but also an estimation of their expected traffic patterns. The reason for this change is that service orchestration is greatly simplified... | {
"cite_N": [
"@cite_12",
"@cite_8"
],
"mid": [
"2612759037",
"2792251914"
],
"abstract": [
"The emerging network slicing paradigm for 5G provides new business opportunities by enabling multi-tenancy support. At the same time, new technical challenges are introduced, as novel resource allo... |
1907.03880 | 2958588938 | When designing swarm-robotic systems, systematic comparison of algorithms from different domains is necessary to determine which is capable of scaling up to handle the target problem size and target operating conditions. We propose a set of quantitative metrics for scalability, flexibility, and emergence which are capa... | In recent years, many theoretical SR system design tools have become available @cite_22 @cite_9 @cite_8 @cite_14 . These tools have made it easier to conduct mathematical analysis of algorithms and derive analytical, rather than weakly inductive proofs of correctness @cite_25 . Despite this, there has not been a corres... | {
"cite_N": [
"@cite_14",
"@cite_26",
"@cite_22",
"@cite_7",
"@cite_8",
"@cite_9",
"@cite_1",
"@cite_19",
"@cite_25",
"@cite_11"
],
"mid": [
"2041478716",
"2114481362",
"2146086743",
"2011995167",
"2274806495",
"2139610651",
"155446141",
"",
... |
1907.03880 | 2958588938 | When designing swarm-robotic systems, systematic comparison of algorithms from different domains is necessary to determine which is capable of scaling up to handle the target problem size and target operating conditions. We propose a set of quantitative metrics for scalability, flexibility, and emergence which are capa... | Within SR, no widely accepted theory of self-organizing systems exists @cite_7 @cite_23 . Cotsaftis presents a control-theoretic model of emergence, distinguishing between systems which can be studied by the methods of scientific reductionism, and systems which originate from the existence of a threshold above which in... | {
"cite_N": [
"@cite_22",
"@cite_7",
"@cite_24",
"@cite_23",
"@cite_2",
"@cite_5",
"@cite_16",
"@cite_13",
"@cite_17"
],
"mid": [
"2146086743",
"2011995167",
"1501224446",
"2120573220",
"",
"2108526989",
"2014403316",
"2107787644",
"280773167... |
1907.03904 | 2968976338 | Blockchains and smart contracts are an emerging, promising technology, that has received considerable attention. We use the blockchain technology, and in particular Ethereum, to implement a large-scale event-based Internet of Things (IoT) control system. We argue that the distributed nature of the “ledger,” as well as,... | Early attempts to incorporate blockchain technology into the IoT proposed new blockchain systems. For example, @cite_1 designed a blockchain-based smart home management system. They proposed a custom, blockchain technology, where the home gateways hold the role of the miners. Such solutions are hard to be deployed sinc... | {
"cite_N": [
"@cite_1"
],
"mid": [
"2611626082"
],
"abstract": [
"Internet of Things (IoT) security and privacy remain a major challenge, mainly due to the massive scale and distributed nature of IoT networks. Blockchain-based approaches provide decentralized security and privacy, yet they involv... |
1907.03904 | 2968976338 | Blockchains and smart contracts are an emerging, promising technology, that has received considerable attention. We use the blockchain technology, and in particular Ethereum, to implement a large-scale event-based Internet of Things (IoT) control system. We argue that the distributed nature of the “ledger,” as well as,... | Recently, @cite_5 explored the potential of smart contracts for machine-to-machine (M2M) communication. To this end, they developed and evaluated an IoT application for automated, M2M, gasoline purchases that uses Ethereum smart contracts to perform transactions. Our work is also in this direction. Nevertheless, in add... | {
"cite_N": [
"@cite_5"
],
"mid": [
"2962852903"
],
"abstract": [
"Blockchain technologies, such as smart contracts, present a unique interface for machine-to-machine communication that provides a secure, append-only record that can be shared without trust and without a central administrator. We s... |
1907.03843 | 2960728668 | The rise of artificial intelligence (A.I.) based systems has the potential to benefit adopters and society as a whole. However, these systems may also enclose potential conflicts and unintended consequences. Notably, people will only adopt an A.I. system if it confers them an advantage, at which point non-adopters migh... | One major problem for the introduction of safe A.I. systems is the so called value alignment problem @cite_37 : How can A.I. systems ensure that their behaviour aligns to the values of their owners? Even though it is not yet a solved problem, for this paper, we will assume that an A.I. system can accurately estimate th... | {
"cite_N": [
"@cite_37"
],
"mid": [
"88368075"
],
"abstract": [
"The principal-agent problem concerns delegation in the absence of trust. Given a principal and an agent with different value structures, the principal wants to motivate the agent to address the principal’s aims by providing appropri... |
1907.03843 | 2960728668 | The rise of artificial intelligence (A.I.) based systems has the potential to benefit adopters and society as a whole. However, these systems may also enclose potential conflicts and unintended consequences. Notably, people will only adopt an A.I. system if it confers them an advantage, at which point non-adopters migh... | A comparison of a number of such ethical frameworks can be found on the paper ''An Ethical Framework for a Good AI Society: Opportunities, Risks, Principles, and Recommendations'' @cite_43 . That paper analyses principles proposed by 6 different entities, including the previously mentioned Asilomar AI principles @cite_... | {
"cite_N": [
"@cite_41",
"@cite_43",
"@cite_42"
],
"mid": [
"",
"2902634493",
"2884040483"
],
"abstract": [
"",
"This article reports the findings of AI4People, an Atomium—EISMD initiative designed to lay the foundations for a “Good AI Society”. We introduce the core opportuni... |
1907.03843 | 2960728668 | The rise of artificial intelligence (A.I.) based systems has the potential to benefit adopters and society as a whole. However, these systems may also enclose potential conflicts and unintended consequences. Notably, people will only adopt an A.I. system if it confers them an advantage, at which point non-adopters migh... | The paper ''Machine Ethics: Creating an Ethical Intelligent Agent'' @cite_22 defends that it may be possible to incorporate an explicit ethical component into a machine relying on inductive logic programming approach. The goal is to solve ethical dilemmas by finding ethical principles that best fit given positive and n... | {
"cite_N": [
"@cite_9",
"@cite_22",
"@cite_11"
],
"mid": [
"2001771035",
"2133460105",
"2051031454"
],
"abstract": [
"",
"The newly emerging field of machine ethics (Anderson and Anderson 2006) is concerned with adding an ethical dimension to machines. Unlike computer ethics -... |
1907.03843 | 2960728668 | The rise of artificial intelligence (A.I.) based systems has the potential to benefit adopters and society as a whole. However, these systems may also enclose potential conflicts and unintended consequences. Notably, people will only adopt an A.I. system if it confers them an advantage, at which point non-adopters migh... | However, some argue that having an ethical framework or even A.I. systems that pass the comparative moral Turing test is not enough @cite_6 . Roman Yampolskiy defends that it is insufficient to have a human-like morality on A.I. systems with super-human intelligence. On such agents, small moral mistakes, common in huma... | {
"cite_N": [
"@cite_7",
"@cite_8",
"@cite_32",
"@cite_6",
"@cite_23",
"@cite_31",
"@cite_12"
],
"mid": [
"2215775476",
"",
"1557461560",
"2238205855",
"2022546212",
"",
"2737838988"
],
"abstract": [
"What happens when machines become more intelligen... |
1907.03843 | 2960728668 | The rise of artificial intelligence (A.I.) based systems has the potential to benefit adopters and society as a whole. However, these systems may also enclose potential conflicts and unintended consequences. Notably, people will only adopt an A.I. system if it confers them an advantage, at which point non-adopters migh... | The focus of most previous works was on considering high-level ethical principles for A.I. systems acting in a society. In most cases there was no claim or prediction about the potential adoption of A.I. systems or their acceptance by non-adopters and by society in general. Just a few works considered the development o... | {
"cite_N": [
"@cite_46"
],
"mid": [
"1012910110"
],
"abstract": [
"This paper presents a simple model of an AI (artificial intelligence) arms race, where several development teams race to build the first AI. Under the assumption that the first AI will be very powerful and transformative, each tea... |
1812.11779 | 2949427863 | Video streaming currently accounts for the majority of Internet traffic. One factor that enables video streaming is HTTP Adaptive Streaming (HAS), that allows the users to stream video using a bit rate that closely matches the available bandwidth from the server to the client. MPEG Dynamic Adaptive Streaming over HTTP ... | The Adaptation Algorithm for Adaptive Streaming over HTTP (AAASH) @cite_7 tries to optimize the user's experience by balancing the following goals: a) to prevent video playback interruptions when possible; b) to maintain a high average and minimum video resolution; c) to decrease the number of resolution changes; d) to... | {
"cite_N": [
"@cite_7"
],
"mid": [
"1967330816"
],
"abstract": [
"Internet video makes up a significant part of the Internet traffic and its fraction is constantly growing. In order to guarantee best user experience throughout different network access technologies with dynamically varying network... |
1812.11779 | 2949427863 | Video streaming currently accounts for the majority of Internet traffic. One factor that enables video streaming is HTTP Adaptive Streaming (HAS), that allows the users to stream video using a bit rate that closely matches the available bandwidth from the server to the client. MPEG Dynamic Adaptive Streaming over HTTP ... | Reference @cite_15 presents an adaptation scheme called Rate Adaptation for Adaptive HTTP Streaming (RAAHS). RAAHS exploits the segment fetch time and compares it to the client's playback time in order to calculate the bit rate of the following segment. Switch up is done using a step-wise process, whereas switch down i... | {
"cite_N": [
"@cite_15"
],
"mid": [
"2061700262"
],
"abstract": [
"Recently, HTTP has been widely used for the delivery of real-time multimedia content over the Internet, such as in video streaming applications. To combat the varying network resources of the Internet, rate adaptation is used to a... |
1812.11779 | 2949427863 | Video streaming currently accounts for the majority of Internet traffic. One factor that enables video streaming is HTTP Adaptive Streaming (HAS), that allows the users to stream video using a bit rate that closely matches the available bandwidth from the server to the client. MPEG Dynamic Adaptive Streaming over HTTP ... | The agile Smooth Video Adaptation Algorithm (SVAA) for DASH systems, proposed in @cite_16 , uses client-side buffered video time as feedback signal to estimate the video rate of the next downloadable video segment. The algorithm increases smoothly the video rate with the available network bandwidth, and it reduces prom... | {
"cite_N": [
"@cite_16"
],
"mid": [
"2157394357"
],
"abstract": [
"Dynamic Adaptive Streaming over HTTP (DASH) is widely deployed on the Internet for live and on-demand video streaming services. Video adaptation algorithms in existing DASH systems are either too sluggish to respond to congestion ... |
1812.11779 | 2949427863 | Video streaming currently accounts for the majority of Internet traffic. One factor that enables video streaming is HTTP Adaptive Streaming (HAS), that allows the users to stream video using a bit rate that closely matches the available bandwidth from the server to the client. MPEG Dynamic Adaptive Streaming over HTTP ... | The authors in @cite_17 replaced the original quality adaption algorithm in Adobe's Open Source Media Framework (OSMF) so that the quality level switching follows a pre-defined scenario. The fetch times of last two video segments are used to estimate the bandwidth that is available between the server and the client. Th... | {
"cite_N": [
"@cite_17"
],
"mid": [
"2104644670"
],
"abstract": [
"Dynamic Adaptation Streaming over HTTP (DASH) enhances the Quality of Experience (QoE) for users by automatically switching quality levels according to network conditions. Various adaptation schemes have been proposed to select th... |
1812.11779 | 2949427863 | Video streaming currently accounts for the majority of Internet traffic. One factor that enables video streaming is HTTP Adaptive Streaming (HAS), that allows the users to stream video using a bit rate that closely matches the available bandwidth from the server to the client. MPEG Dynamic Adaptive Streaming over HTTP ... | @cite_5 , the authors applied a Markov chain to analyse the QoE metrics for the user, namely the probability of the video to be interrupted, the initial buffering delay, the average bit rate of the video, and the rate of bit rate changes. | {
"cite_N": [
"@cite_5"
],
"mid": [
"2804849494"
],
"abstract": [
"Adaptive video streaming improves users' quality of experience (QoE), while using the network efficiently. In the last few years, adaptive video streaming has seen widespread adoption and has attracted significant research effort. ... |
1812.11779 | 2949427863 | Video streaming currently accounts for the majority of Internet traffic. One factor that enables video streaming is HTTP Adaptive Streaming (HAS), that allows the users to stream video using a bit rate that closely matches the available bandwidth from the server to the client. MPEG Dynamic Adaptive Streaming over HTTP ... | The work in @cite_10 @cite_11 takes an advantage of a fuzzy logic application to handle the uncertainty of the network system. In @cite_11 , the mobile QoS is improved by using a cumulative moving average in order to capture the related information between near-term past values and current values. In order to enhance u... | {
"cite_N": [
"@cite_18",
"@cite_14",
"@cite_1",
"@cite_10",
"@cite_11"
],
"mid": [
"2508578164",
"2805947310",
"2744786564",
"2601347938",
"2750827144"
],
"abstract": [
"In this paper, we present a Mobile Edge Computing (MEC) scheme for enabling network edge-assist... |
1812.11779 | 2949427863 | Video streaming currently accounts for the majority of Internet traffic. One factor that enables video streaming is HTTP Adaptive Streaming (HAS), that allows the users to stream video using a bit rate that closely matches the available bandwidth from the server to the client. MPEG Dynamic Adaptive Streaming over HTTP ... | Recent surveys @cite_4 @cite_3 give a good overview of the bit rate adaptation algorithms for DASH based content delivery. | {
"cite_N": [
"@cite_4",
"@cite_3"
],
"mid": [
"2602023803",
"2906364736"
],
"abstract": [
"With companies such as Netflix and YouTube accounting for more than 50 of the peak download traffic on North American fixed networks in 2015, video streaming represents a significant source of Inter... |
1812.11941 | 2907670226 | Accurate depth estimation from images is a fundamental task in many applications including scene understanding and reconstruction. Existing solutions for depth estimation often produce blurry approximations of low resolution. This paper presents a convolutional neural network for computing a high-resolution depth map g... | has been considered by many CNN methods where they formulate the problem as a regression of the depth map from a single RGB image @cite_2 @cite_18 @cite_14 @cite_6 @cite_26 @cite_36 . While the performance of these methods have been increasing steadily, general problems in both the quality and resolution of the estimat... | {
"cite_N": [
"@cite_18",
"@cite_14",
"@cite_26",
"@cite_36",
"@cite_6",
"@cite_2"
],
"mid": [
"2963591054",
"2605938684",
"2964014680",
"2963488291",
"2890173472",
"2171740948"
],
"abstract": [
"This paper addresses the problem of estimating the depth map o... |
1812.11941 | 2907670226 | Accurate depth estimation from images is a fundamental task in many applications including scene understanding and reconstruction. Existing solutions for depth estimation often produce blurry approximations of low resolution. This paper presents a convolutional neural network for computing a high-resolution depth map g... | stereo reconstruction using CNN algorithms have been recently proposed @cite_40 . Prior work considered the subproblem that looks at image pairs @cite_17 , or three consecutive frames @cite_35 . Joint key-frame based dense camera tracking and depth map estimation was presented by @cite_20 . In this work, we seek to pus... | {
"cite_N": [
"@cite_35",
"@cite_40",
"@cite_20",
"@cite_17"
],
"mid": [
"2806446538",
"2964153986",
"2887825894",
"2561074213"
],
"abstract": [
"Per-pixel ground-truth depth data is challenging to acquire at scale. To overcome this limitation, self-supervised learning has ... |
1812.11941 | 2907670226 | Accurate depth estimation from images is a fundamental task in many applications including scene understanding and reconstruction. Existing solutions for depth estimation often produce blurry approximations of low resolution. This paper presents a convolutional neural network for computing a high-resolution depth map g... | approaches have been shown to be very helpful in many different contexts. In recent work, investigated the efficiency of transfer learning between different tasks @cite_3 , many of which were are related to 3D reconstruction. Our method is heavily based on the idea of transfer learning where we make use of image encode... | {
"cite_N": [
"@cite_34",
"@cite_3"
],
"mid": [
"2963446712",
"2964185501"
],
"abstract": [
"Recent work has shown that convolutional networks can be substantially deeper, more accurate, and efficient to train if they contain shorter connections between layers close to the input and those ... |
1812.11941 | 2907670226 | Accurate depth estimation from images is a fundamental task in many applications including scene understanding and reconstruction. Existing solutions for depth estimation often produce blurry approximations of low resolution. This paper presents a convolutional neural network for computing a high-resolution depth map g... | networks have made significant contributions in many vision related problems such as image segmentation @cite_10 , optical flow estimation @cite_25 , and image restoration @cite_27 . In recent years, the use of such architectures have shown great success both in the supervised and the unsupervised setting of the depth ... | {
"cite_N": [
"@cite_8",
"@cite_27",
"@cite_40",
"@cite_10",
"@cite_25",
"@cite_20",
"@cite_17"
],
"mid": [
"2520707372",
"2964204553",
"2964153986",
"1901129140",
"764651262",
"2887825894",
"2561074213"
],
"abstract": [
"Learning based methods have ... |
1812.11671 | 2907508336 | At present, deep learning has been applied more and more in monocular image depth estimation and has shown promising results. The current more ideal method for monocular depth estimation is the supervised learning based on ground truth depth, but this method requires an abundance of expensive ground truth depth as the ... | Due to the rise of robotics and virtual reality, depth evaluation has undoubtedly become one of the most popular research points at present. Because machine learning or deep learning has better performance than traditional methods, more and more researchers has applied this methods to depth evaluation and some research... | {
"cite_N": [
"@cite_27",
"@cite_31"
],
"mid": [
"2440384215",
"2139905387"
],
"abstract": [
"In the past year, convolutional neural networks have been shown to perform extremely well for stereo estimation. However, current architectures rely on siamese networks which exploit concatenation... |
1812.11671 | 2907508336 | At present, deep learning has been applied more and more in monocular image depth estimation and has shown promising results. The current more ideal method for monocular depth estimation is the supervised learning based on ground truth depth, but this method requires an abundance of expensive ground truth depth as the ... | We can see the comparison results from Table , that the stereo matching models outperforms monocular depth estimation models under the same unsupervised depth estimation model. So inspired by @cite_11 , we proposed a unsupervised monocular image stereo matching model that composed by the view synthesis network and ster... | {
"cite_N": [
"@cite_11"
],
"mid": [
"2794293902"
],
"abstract": [
"Previous monocular depth estimation methods take a single view and directly regress the expected results. Though recent advances are made by applying geometrically inspired loss functions during training, the inference procedure d... |
1812.11834 | 2907413770 | Face restoration from low resolution and noise is important for applications of face analysis recognition. However, most existing face restoration models omit the multiple scale issues in face restoration problem, which is still not well-solved in research area. In this paper, we propose a Sequential Gating Ensemble Ne... | face restoration is of great importance for vision applications. Therefore, extensive studies have been carried out to restore the low quality face image to high quality face image in the past decades. The early face restoration algorithms can be categorized into two classes, i.e., global face-based restoration methods... | {
"cite_N": [
"@cite_37",
"@cite_0",
"@cite_27",
"@cite_10",
"@cite_17"
],
"mid": [
"",
"2121058967",
"2015497428",
"2103871101",
"2171107009"
],
"abstract": [
"",
"This paper presents a new approach to single-image superresolution, based upon sparse signal repr... |
1812.11834 | 2907413770 | Face restoration from low resolution and noise is important for applications of face analysis recognition. However, most existing face restoration models omit the multiple scale issues in face restoration problem, which is still not well-solved in research area. In this paper, we propose a Sequential Gating Ensemble Ne... | To overcome the drawback of global face-based restoration methods, local patch-based restoration methods decompose face image into small patches, which can capture more facial details. Local patch-based restoration methods assume that LR and HR face patch manifolds are locally isometric. Therefore, once obtaining the r... | {
"cite_N": [
"@cite_36",
"@cite_35",
"@cite_14",
"@cite_22"
],
"mid": [
"2141631520",
"2509704168",
"2027325144",
"2031349574"
],
"abstract": [
"In video surveillance, the faces of interest are often of small size. Image resolution is an important factor affecting face rec... |
1812.11834 | 2907413770 | Face restoration from low resolution and noise is important for applications of face analysis recognition. However, most existing face restoration models omit the multiple scale issues in face restoration problem, which is still not well-solved in research area. In this paper, we propose a Sequential Gating Ensemble Ne... | In the past few years, convolutional neural networks (CNN) @cite_29 have shown an explosive popularity and success in various computer vision fields, such as image recognition @cite_6 , object detection @cite_3 , face recognition @cite_40 , and semantic segmentation @cite_32 . CNN based image restoration algorithms hav... | {
"cite_N": [
"@cite_18",
"@cite_33",
"@cite_28",
"@cite_29",
"@cite_32",
"@cite_3",
"@cite_6",
"@cite_40",
"@cite_2",
"@cite_34"
],
"mid": [
"2963410064",
"54257720",
"2523714292",
"2147800946",
"1903029394",
"2953106684",
"2194775991",
"232... |
1812.11834 | 2907413770 | Face restoration from low resolution and noise is important for applications of face analysis recognition. However, most existing face restoration models omit the multiple scale issues in face restoration problem, which is still not well-solved in research area. In this paper, we propose a Sequential Gating Ensemble Ne... | However, these skip-connections in @cite_28 @cite_34 fail to explore the underlying sequential relationship among multi-level feature maps in image restoration problem. Therefore, we design our SGEN followed by the goal of autoencoder, which sequentially extracts high level information from base-encoders in bottom-up m... | {
"cite_N": [
"@cite_28",
"@cite_34"
],
"mid": [
"2523714292",
"2964046669"
],
"abstract": [
"Despite the breakthroughs in accuracy and speed of single image super-resolution using faster and deeper convolutional neural networks, one central problem remains largely unsolved: how do we reco... |
1812.11741 | 2907694156 | We consider multi-agent systems where agents actions and beliefs are determined aleatorically, or "by the throw of dice". This system consists of possible worlds that assign distributions to independent random variables, and agents who assign probabilities to these possible worlds. We present a novel syntax and semanti... | These approaches lose the simplicity of Boolean logics, as deductive systems must deal with propositions that are not independent. This limits their practicality as well defined semantics require the conditional probabilities of all atoms to be known. However, these approaches have been successfully combined with logic... | {
"cite_N": [
"@cite_14",
"@cite_10",
"@cite_6",
"@cite_12"
],
"mid": [
"2738790068",
"",
"2000108089",
"2400548963"
],
"abstract": [
"We present an implementation of a probabilistic first-order logic called TensorLog, in which classes of logical queries are compiled into d... |
1812.11741 | 2907694156 | We consider multi-agent systems where agents actions and beliefs are determined aleatorically, or "by the throw of dice". This system consists of possible worlds that assign distributions to independent random variables, and agents who assign probabilities to these possible worlds. We present a novel syntax and semanti... | More general foundational work on reasoning probabilistically was done by de Finetti @cite_3 who established an epistemic notion of probability based on what an agent would consider to be a rational wager (the Dutch book argument). In @cite_21 , Milne incorporates these ideas into the logic of conditional events. Staln... | {
"cite_N": [
"@cite_9",
"@cite_21",
"@cite_3",
"@cite_17"
],
"mid": [
"1590902735",
"1982537065",
"2172298389",
""
],
"abstract": [
"A conditional sentence expresses a proposition which is a function of two other propositions, yet not one which is a truth function of those... |
1812.11321 | 2949347774 | A capsule is a group of neurons, whose activity vector represents the instantiation parameters of a specific type of entity. In this paper, we explore the capsule networks used for relation extraction in a multi-instance multi-label learning framework and propose a novel neural approach based on capsule networks with a... | In the recent years, NN models have shown superior performance over approaches using hand-crafted features in various tasks. CNN is the first one of the deep learning models that have been applied to relation extraction @cite_14 . Variants of convolutional networks include piecewise-CNN (PCNN) @cite_16 , instance-level... | {
"cite_N": [
"@cite_14",
"@cite_9",
"@cite_6",
"@cite_3",
"@cite_0",
"@cite_23",
"@cite_15",
"@cite_16",
"@cite_12",
"@cite_11"
],
"mid": [
"2950371387",
"2515462165",
"2783378231",
"",
"2787051437",
"",
"2963912690",
"2251135946",
"2612... |
1812.11321 | 2949347774 | A capsule is a group of neurons, whose activity vector represents the instantiation parameters of a specific type of entity. In this paper, we explore the capsule networks used for relation extraction in a multi-instance multi-label learning framework and propose a novel neural approach based on capsule networks with a... | Recently, the capsule network has been proposed to improve the representation limitations of CNNs and RNNs. @cite_13 replaced the scalar-output feature detectors of CNNs with vector-output capsules and max-pooling with routing-by-agreement. @cite_21 ) proposed a new iterative routing procedure among capsule layers, bas... | {
"cite_N": [
"@cite_21",
"@cite_1",
"@cite_19",
"@cite_13",
"@cite_20"
],
"mid": [
"2785994986",
"2796138868",
"2805853672",
"2963703618",
"2788347302"
],
"abstract": [
"A capsule is a group of neurons whose outputs represent different properties of the same entity... |
1812.11252 | 2908253384 | As science advances, the academic community has published millions of research papers. Researchers devote time and effort to search relevant manuscripts when writing a paper or simply to keep up with current research. In this paper, we consider the problem of citation recommendation by extending a set of known-to-be-re... | Given a "basket" of citations, @cite_31 explore the use of Collaborative Filtering (CF) to recommend papers that would be suitable additional references for a target research paper. They create a ratings matrix where citing papers correspond to users and citations correspond to items. The experiments show CF could gene... | {
"cite_N": [
"@cite_19",
"@cite_31",
"@cite_16"
],
"mid": [
"2139375986",
"2116655493",
"2142574815"
],
"abstract": [
"All new researchers face the daunting task of familiarizing themselves with the existing body of research literature in their respective fields. Recommender algor... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.