source stringlengths 389 24.7k | summary stringlengths 92 1.61k | source_num_tokens int64 88 5.16k | summary_num_tokens int64 26 445 |
|---|---|---|---|
In this paper we propose a novel easily reproducible technique to attack the best public Face ID system ArcFace in different shooting conditions. To create an attack, we print the rectangular paper sticker on a common color printer and put it on the hat. The adversarial sticker is prepared with a novel algorithm for of... | Another interesting approach @cite_14 uses the concept of nested adversarial examples where separate non-overlapping adversarial perturbations are generated for close and far distances. This attack is designed for Faster R-CNN and YOLOv3 @cite_38 . | 570 | 62 |
In this paper we propose a novel easily reproducible technique to attack the best public Face ID system ArcFace in different shooting conditions. To create an attack, we print the rectangular paper sticker on a common color printer and put it on the hat. The adversarial sticker is prepared with a novel algorithm for of... | A few works are devoted to more complex approaches. One of such works @cite_29 proposes to use EOT, NPS, and TV loss for fooling YOLOv2-based person detector. Another one @cite_2 is devoted to fooling the Face ID system using adversarial generative nets (a sort of GANs @cite_18 ) where the generator produces the eyegla... | 854 | 96 |
For decades, the join operator over fast data streams has always drawn much attention from the database community, due to its wide spectrum of real-world applications, such as online clustering, intrusion detection, sensor data monitoring, and so on. Existing works usually assume that the underlying streams to be joine... | Stream Processing. There are many important problems for stream data processing, including event detection @cite_35 , outlier detection @cite_16 , top- @math query @cite_37 , join @cite_38 @cite_22 , skyline query @cite_12 , nearest neighbor query @cite_33 , aggregate query @cite_17 , and so on. These works usually ass... | 2,158 | 126 |
For decades, the join operator over fast data streams has always drawn much attention from the database community, due to its wide spectrum of real-world applications, such as online clustering, intrusion detection, sensor data monitoring, and so on. Existing works usually assume that the underlying streams to be joine... | Differential Dependency. Differential dependency (DD) @cite_18 is a valuable tool for data imputation @cite_2 , data cleaning @cite_14 , data repairing @cite_39 , and so on. @cite_2 used the DDs to fill the missing attributes of incomplete objects on static data set via some detected neighbors satisfying the distance c... | 1,724 | 159 |
For decades, the join operator over fast data streams has always drawn much attention from the database community, due to its wide spectrum of real-world applications, such as online clustering, intrusion detection, sensor data monitoring, and so on. Existing works usually assume that the underlying streams to be joine... | Join Over Certain Uncertain Databases. The join operator was traditionally used in relational databases @cite_15 or data streams @cite_38 . The join predicate may follow equality semantics between attributes of tuples or data objects. According to predicate constraints, join over uncertain databases @cite_11 @cite_22 c... | 1,258 | 135 |
For decades, the join operator over fast data streams has always drawn much attention from the database community, due to its wide spectrum of real-world applications, such as online clustering, intrusion detection, sensor data monitoring, and so on. Existing works usually assume that the underlying streams to be joine... | PSJ has received much attention in many domains. @cite_19 applied PSJ to integrate heterogeneous RDF graphs by introducing an equivalent semantics for RDF graphs. @cite_31 proposed an effective filter-based method for high-dimensional vector similarity join. @cite_30 explored how to leverage relations between sets to p... | 1,577 | 225 |
For decades, the join operator over fast data streams has always drawn much attention from the database community, due to its wide spectrum of real-world applications, such as online clustering, intrusion detection, sensor data monitoring, and so on. Existing works usually assume that the underlying streams to be joine... | Incomplete Databases. In the literature of incomplete databases, the most commonly used imputation methods include rule-based @cite_1 , statistical-based @cite_36 , pattern-based @cite_20 , constraint-based @cite_42 imputation, and so on. These existing works may incur the accuracy problem for sparse data sets. That is... | 1,637 | 197 |
Neural plasticity is an important functionality of human brain, in which number of neurons and synapses can shrink or expand in response to stimuli throughout the span of life. We model this dynamic learning process as an @math -norm regularized binary optimization problem, in which each unit of a neural network (e.g.,... | Another closely related area is neural architecture search @cite_24 @cite_29 @cite_6 that searches for an optimal network architecture for a given learning task. It attempts to determine number of layers, types of layers, layer configurations, different activation functions, etc. Given the extremely large search space,... | 1,191 | 130 |
Neural plasticity is an important functionality of human brain, in which number of neurons and synapses can shrink or expand in response to stimuli throughout the span of life. We model this dynamic learning process as an @math -norm regularized binary optimization problem, in which each unit of a neural network (e.g.,... | Compared to network sparsification, network expansion is a relatively less explored area. There are few existing works that can dynamically increase the capacity of network during training. For example, DNC @cite_16 sequentially adds neurons one at a time to the hidden layers of network until the desired approximation ... | 1,080 | 229 |
Dynamic network slicing has emerged as a promising and fundamental framework for meeting 5G's diverse use cases. As machine learning (ML) is expected to play a pivotal role in the efficient control and management of these networks, in this work we examine the ML-based Quality-of-Transmission (QoT) estimation problem un... | Several ML applications have been already developed and explored for optical network planning purposes @cite_9 @cite_8 . In general, the state-of-the-art assumes that the optical network is centrally controlled by an SDN-based optical network controller @cite_11 @cite_20 , equipped with storage, processing, and monitor... | 1,067 | 162 |
Existing deep learning approaches on 3d human pose estimation for videos are either based on Recurrent or Convolutional Neural Networks (RNNs or CNNs). However, RNN-based frameworks can only tackle sequences with limited frames because sequential models are sensitive to bad frames and tend to drift over long sequences.... | The inherent depth ambiguity in 3d pose estimation from monocular images limits the estimation accuracy. Extensive research has been done to exploit extra information contained in temporal sequences. Zhou al @cite_35 formulate an optimization problem to search for the 3d configuration with the highest probability given... | 1,025 | 169 |
Existing deep learning approaches on 3d human pose estimation for videos are either based on Recurrent or Convolutional Neural Networks (RNNs or CNNs). However, RNN-based frameworks can only tackle sequences with limited frames because sequential models are sensitive to bad frames and tend to drift over long sequences.... | Recently, RNN-based frameworks are used to deal with sequential input data. Lin al @cite_1 use a multi-stage framework based on Long Short-term Memory (LSTM) units to estimate the 3d pose from the extracted 2d features and estimated 3d pose in the previous stage. Coskun al @cite_32 propose to learn a human motion model... | 1,450 | 255 |
Existing deep learning approaches on 3d human pose estimation for videos are either based on Recurrent or Convolutional Neural Networks (RNNs or CNNs). However, RNN-based frameworks can only tackle sequences with limited frames because sequential models are sensitive to bad frames and tend to drift over long sequences.... | Inspired by matrix factorization methods commonly used in Structure-from-Motion (SfM) @cite_23 and non-rigid SfM @cite_33 , several works @cite_3 @cite_45 @cite_21 on 3d human pose estimation factorize the sequence of 3d human poses into a linear combination of shape bases. Akhter al @cite_42 suggest a duality of the f... | 1,320 | 161 |
Among various optimization algorithms, ADAM can achieve outstanding performance and has been widely used in model learning. ADAM has the advantages of fast convergence with both momentum and adaptive learning rate. For deep neural network learning problems, since their objective functions are nonconvex, ADAM can also g... | : Deep learning models have achieved great success in recent years, whose representative examples include convolutional neural network(CNN) @cite_12 @cite_1 . CNN has been mainly used to deal with image data and shows outstanding performance on various computer vision tasks. Besides CNN, there also exist many other typ... | 1,319 | 128 |
Among various optimization algorithms, ADAM can achieve outstanding performance and has been widely used in model learning. ADAM has the advantages of fast convergence with both momentum and adaptive learning rate. For deep neural network learning problems, since their objective functions are nonconvex, ADAM can also g... | : Boosting was proposed by @cite_8 . Given a training dataset containing @math training examples, a batch of @math training examples is generated by random sampling with replacement. We can generate @math training sets from the same original training data by applying the sampling @math times @cite_20 , which will be us... | 656 | 90 |
Recognizing the layout of unstructured digital documents is an important step when parsing the documents into structured machine-readable format for downstream applications. Deep neural networks that are developed for computer vision have been proven to be an effective method to analyze layout of document images. Howev... | Existing datasets for document layout analysis rely on manual annotation. Some of these datasets are used in document processing challenges. Examples of these efforts are available in several ICDAR challenges @cite_19 , which cover as well complex layouts @cite_18 @cite_7 . The US NIH National Library of Medicine has p... | 819 | 138 |
Recognizing the layout of unstructured digital documents is an important step when parsing the documents into structured machine-readable format for downstream applications. Deep neural networks that are developed for computer vision have been proven to be an effective method to analyze layout of document images. Howev... | In addition to document layout, further understanding of the document content has been studied in the evaluation of table detection methods, e.g. @cite_13 @cite_21 . Examples include table detection from document images using heuristics @cite_11 , vertical arrangement of text blocks @cite_3 and deep learning methods @c... | 2,034 | 89 |
Instance segmentation requires a large number of training samples to achieve satisfactory performance and benefits from proper data augmentation. To enlarge the training set and increase the diversity, previous methods have investigated using data annotation from other domain (e.g. bbox, point) in a weakly supervised m... | Combining instance detection and semantic segmentation, instance segmentation @cite_45 @cite_19 @cite_4 @cite_29 @cite_39 @cite_5 @cite_11 @cite_27 is a much harder problem. Earlier methods either propose segmentation candidates followed by classification @cite_40 , or associate pixels on the semantic segmentation map ... | 2,273 | 293 |
Multi-Task Learning (MTL) aims at boosting the overall performance of each individual task by leveraging useful information contained in multiple related tasks. It has shown great success in natural language processing (NLP). Currently, a number of MLT architectures and learning mechanisms have been proposed for variou... | Multi-task learning with deep neural networks has gained increasing attention within NLP community over the past decades. @cite_3 and @cite_21 described most of the existing techniques for multi-task learning in deep neural networks. Generally, existing MTL methods can be categorised as @cite_29 @cite_10 and @cite_20 @... | 1,092 | 79 |
Clinicians and other analysts working with healthcare data are in need for better support to cope with large and complex data. While an increasing number of visual analytics environments integrates explicit domain knowledge as a means to deliver a precise representation of the available data, theoretical work so far ha... | Since Illuminating the Path'' [p. ,35] thomas_2005_illuminating , incorporating prior domain knowledge and build[ing] knowledge structures'' has been on VA's agenda. This is underscored by the pivotal position of knowledge in the VA process model by @cite_40 @cite_36 and further process models such as the knowledge gen... | 1,310 | 205 |
Clinicians and other analysts working with healthcare data are in need for better support to cope with large and complex data. While an increasing number of visual analytics environments integrates explicit domain knowledge as a means to deliver a precise representation of the available data, theoretical work so far ha... | Beyond the role of knowledge in the VA process, only few works discuss the content and structure of explicit knowledge on a general level. @cite_7 conceptualize domain knowledge as a model of a part of reality and provide definitions for different types of models but they do not specify the form and medium how the mode... | 1,304 | 275 |
Clinicians and other analysts working with healthcare data are in need for better support to cope with large and complex data. While an increasing number of visual analytics environments integrates explicit domain knowledge as a means to deliver a precise representation of the available data, theoretical work so far ha... | , it can be seen, that most of the discussed approaches cover how explicit domain knowledge can be exploited to enhance visual representation and data analysis; some approaches provide methods to generate explicit knowledge. Additionally, most of the currently implemented knowledge-assisted VA environments are focused ... | 350 | 171 |
Neural network image reconstruction directly from measurement data is a growing field of research, but until now has been limited to producing small (e.g. 128x128) 2D images by the large memory requirements of the previously suggested networks. In order to facilitate further research with direct reconstruction, we deve... | The terms deep learning and image reconstruction are often used in conjunction to describe a significant amount of recent research @cite_24 that most often falls into one of two categories: 1) combine deep learning with an analytical or statistical method such as using a deep learning prior @cite_13 or regularization t... | 1,340 | 124 |
Neural network image reconstruction directly from measurement data is a growing field of research, but until now has been limited to producing small (e.g. 128x128) 2D images by the large memory requirements of the previously suggested networks. In order to facilitate further research with direct reconstruction, we deve... | Early research in this area was based on networks of fully connected multilayer perceptrons @cite_11 @cite_12 @cite_17 @cite_14 that yielded promising results, but only for very low resolution reconstructions. More recent efforts have capitalized on the growth of computational resources, especially in the area of GPUs,... | 1,155 | 295 |
Dialogue state tracking (DST) is an essential component in task-oriented dialogue systems, which estimates user goals at every dialogue turn. However, most previous approaches usually suffer from the following problems. Many discriminative models, especially end-to-end (E2E) models, are difficult to extract unknown val... | As far as we know, @cite_4 is the only work using the discriminative model to handle the dynamic and unbounded value set. @cite_4 represents the dialogue states by candidate sets derived from the dialogue and knowledge, then scores values in the candidate set with binary classifications. Although the sophisticated gene... | 597 | 96 |
A significantly low cost and tractable progressive learning approach is proposed and discussed for efficient spatiotemporal monitoring of a completely unknown, two dimensional correlated signal distribution in localized wireless sensor field. The spatial distribution is compressed into a number of its contour lines and... | Contour line detection in wireless sensor networks, which is the first step in modeling the spatial distribution has been addressed in several researches, including @cite_13 @cite_31 @cite_1 @cite_17 @cite_18 @cite_15 @cite_26 @cite_30 @cite_12 @cite_20 @cite_16 . Most of these mentioned researches addressed to distrib... | 1,416 | 136 |
A significantly low cost and tractable progressive learning approach is proposed and discussed for efficient spatiotemporal monitoring of a completely unknown, two dimensional correlated signal distribution in localized wireless sensor field. The spatial distribution is compressed into a number of its contour lines and... | Spatial modeling of signal distributions using contour lines has been addressed in @cite_13 @cite_31 @cite_21 . Modeling the spatial distribution with uniformly spaced contour levels and tracking their variation using time-series analysis in sensors was studied in @cite_21 . Using non-uniformly spaced contour lines was... | 657 | 140 |
A significantly low cost and tractable progressive learning approach is proposed and discussed for efficient spatiotemporal monitoring of a completely unknown, two dimensional correlated signal distribution in localized wireless sensor field. The spatial distribution is compressed into a number of its contour lines and... | Spatiotemporal modeling using machine learning approaches has been reported in several researches, including @cite_9 @cite_5 @cite_19 . In most of these approaches, neural networks algorithms, genetics algorithms, stochastic gradient descent algorithms, etc. are employed. | 982 | 61 |
Online platforms, such as Airbnb, this http URL, Amazon, Uber and Lyft, can control and optimize many aspects of product search to improve the efficiency of marketplaces. Here we focus on a common model, called the discriminatory control model, where the platform chooses to display a subset of sellers who sell products... | Bertrand competition, proposed by Joseph Bertrand in 1883, and Cournot competition, introduced in 1838 by Antoine Augustin Cournot, are fundamental economic models that represent sellers competing in a single market, and have been studied comprehensively in economics. Due to the motivation that many sellers compete in ... | 882 | 271 |
The emerging parallel chain protocols represent a breakthrough to address the scalability of blockchain. By composing multiple parallel chain instances, the whole systems' throughput can approach the network capacity. How to coordinate different chains' blocks and to construct them into a global ordering is critical to... | In @cite_39 , Jiaping Wang and Hao Wang propose a protocol called Monoxide, which composes multiple independent single chain consensus systems, called zones. They also proposed eventual atomicity to ensure transaction atomicity across zones and Chu-ko-nu mining to ensure the effective mining power in each zone. Monoxid... | 1,322 | 199 |
Action recognition has seen a dramatic performance improvement in the last few years. Most of the current state-of-the-art literature either aims at improving performance through changes to the backbone CNN network, or they explore different trade-offs between computational efficiency and performance, again through alt... | Action recognition in the deep learning era has been successfully tackled with 2D @cite_12 @cite_1 and 3D CNNs @cite_31 @cite_29 @cite_0 @cite_28 @cite_40 @cite_15 @cite_30 . Most existing works focus on modeling of motion and temporal structures. @cite_12 the optical flow CNN is introduced to model short-term motion p... | 2,223 | 252 |
Action recognition has seen a dramatic performance improvement in the last few years. Most of the current state-of-the-art literature either aims at improving performance through changes to the backbone CNN network, or they explore different trade-offs between computational efficiency and performance, again through alt... | Most of these methods treat action recognition as a video classification problem. These works tend to focus on how motion is captured by the networks and largely ignore what makes the actions unique. In this work, we provide insights specific to the nature of the action recognition problem itself, showing how it requir... | 451 | 144 |
Action recognition has seen a dramatic performance improvement in the last few years. Most of the current state-of-the-art literature either aims at improving performance through changes to the backbone CNN network, or they explore different trade-offs between computational efficiency and performance, again through alt... | Understanding human activities as a fine-grained recognition problem has been explored for some domain specific tasks @cite_6 @cite_37 . For example, some works have been proposed for hand-gesture recognition @cite_18 @cite_38 @cite_32 , daily life activity recognition @cite_39 and sports understanding @cite_26 @cite_2... | 1,776 | 149 |
Action recognition has seen a dramatic performance improvement in the last few years. Most of the current state-of-the-art literature either aims at improving performance through changes to the backbone CNN network, or they explore different trade-offs between computational efficiency and performance, again through alt... | Different from common object categories such as those in ImageNet @cite_21 , this field cares for objects that look visually very similar and that can only be differentiated by learning their finer details. Some examples including distinguishing bird @cite_5 and plant @cite_9 species and recognizing different car model... | 3,291 | 268 |
Transfer learning enhances learning across tasks, by leveraging previously learned representations -- if they are properly chosen. We describe an efficient method to accurately estimate the appropriateness of a previously trained model for use in a new learning task. We use this measure, which we call "Predict To Learn... | The transfer learning literature explores a number of different topics and strategies such as few-shot learning @cite_19 @cite_29 , domain adaptation @cite_6 , weight synthesis @cite_2 , and multi-task learning @cite_33 @cite_24 @cite_8 . Some works propose novel combinations of these approaches, yielding new training ... | 1,554 | 106 |
Transfer learning enhances learning across tasks, by leveraging previously learned representations -- if they are properly chosen. We describe an efficient method to accurately estimate the appropriateness of a previously trained model for use in a new learning task. We use this measure, which we call "Predict To Learn... | Several approaches have been tried to transfer robust representations based on large numbers of examples to new tasks. These transfer learning approaches share a common intuition @cite_28 : that networks which have learned compact representations of a "source" task, can reuse these representations to achieve higher per... | 1,938 | 287 |
Transfer learning enhances learning across tasks, by leveraging previously learned representations -- if they are properly chosen. We describe an efficient method to accurately estimate the appropriateness of a previously trained model for use in a new learning task. We use this measure, which we call "Predict To Learn... | While all these methods seek to improve performance on the target task by transfer from the source task, most assume there is only one source model, usually trained from ImageNet. @cite_15 Additionally, this approach involves a number of meta-learning decisions, although in general each change from the original source ... | 1,380 | 213 |
Transfer learning enhances learning across tasks, by leveraging previously learned representations -- if they are properly chosen. We describe an efficient method to accurately estimate the appropriateness of a previously trained model for use in a new learning task. We use this measure, which we call "Predict To Learn... | Our approach is most similar to that of fine-tuning with co-training @cite_14 . That method begins by using low-level features to identify images within a source dataset having similar textures to a target dataset, and concludes by using a multi-task objective to fine-tune the target task using these images. A related ... | 1,421 | 186 |
Most of the existing generative adversarial networks (GAN) for text generation suffer from the instability of reinforcement learning training algorithms such as policy gradient, leading to unstable performance. To tackle this problem, we propose a novel framework called Adversarial Reward Augmented Maximum Likelihood (... | As mentioned above, MLE suffers from the exposure bias problem @cite_3 @cite_33 . Thus, reinforcement learning has been introduced to text generation tasks such as policy gradient @cite_33 and actor-critic @cite_20 . @cite_37 proposed an efficient and stable approach called Reward Augmented Maximum Likelihood (RAML), w... | 935 | 102 |
Most of the existing generative adversarial networks (GAN) for text generation suffer from the instability of reinforcement learning training algorithms such as policy gradient, leading to unstable performance. To tackle this problem, we propose a novel framework called Adversarial Reward Augmented Maximum Likelihood (... | The most similar works to our model are RAML @cite_37 and MaliGAN @cite_1 : 1) Compared with RAML, our model adds a discriminator to learn the reward signals instead of choosing existing metrics as rewards. We believe that our model can adapt to various text generation tasks, particularly those without explicit evaluat... | 537 | 113 |
Recognizing multiple labels of images is a practical and challenging task, and significant progress has been made by searching semantic-aware regions and modeling label dependency. However, current methods cannot locate the semantic regions accurately due to the lack of part-level supervision or semantic guidance. More... | Recent progress on multi-label image classification relies on the combination of object localization and deep learning techniques @cite_28 @cite_10 . Generally, they introduced object proposals @cite_22 that were assumed to contain all possible foreground objects in the image and aggregated features extracted from all ... | 1,985 | 260 |
RGB-Thermal object tracking attempt to locate target object using complementary visual and thermal infrared data. Existing RGB-T trackers fuse different modalities by robust feature representation learning or adaptive modal weighting. However, how to integrate dual attention mechanism for visual tracking is still a sub... | RGB-T tracking receives more and more attention in computer vision community with the popularity of thermal infrared sensors. Wu @cite_15 concatenate the image patches from RGB and thermal sources, and then sparsely represent each sample in the target template space for tracking. Modal weights are introduced for each s... | 1,120 | 183 |
RGB-Thermal object tracking attempt to locate target object using complementary visual and thermal infrared data. Existing RGB-T trackers fuse different modalities by robust feature representation learning or adaptive modal weighting. However, how to integrate dual attention mechanism for visual tracking is still a sub... | Attention mechanism originates from the study of human congnitive neuroscience @cite_9 . In visual tracking, the cosine window map @cite_8 and Gaussian window map @cite_3 are widely used in DCF tracker to suppress the boundary effect, which can interpreted as one type of visual spatial attention. For short-time trackin... | 1,737 | 173 |
Convolutional Neural Networks (CNN) have become state-of-the-art in the field of image classification. However, not everything is understood about their inner representations. This paper tackles the interpretability and explainability of the predictions of CNNs for multi-class classification problems. Specifically, we ... | Input modification methods observe the changes in the outputs based on changes to the inputs. This can be done using perturbed variants @cite_0 @cite_40 , masks @cite_36 , or noise @cite_1 . For instance, Zeiler and Fergus @cite_36 create heatmaps based on the drop in prediction probability from input images with maske... | 677 | 114 |
Convolutional Neural Networks (CNN) have become state-of-the-art in the field of image classification. However, not everything is understood about their inner representations. This paper tackles the interpretability and explainability of the predictions of CNNs for multi-class classification problems. Specifically, we ... | Back propagation-based methods trace the contribution of the output backwards through the network to the input. In the classic example, DeconvNet @cite_36 @cite_35 back propagates the output through the network. However, instead of using the Rectified Linear Unit (ReLU) from the forward pass, DeconvNet applies ReLU to ... | 569 | 118 |
Convolutional Neural Networks (CNN) have become state-of-the-art in the field of image classification. However, not everything is understood about their inner representations. This paper tackles the interpretability and explainability of the predictions of CNNs for multi-class classification problems. Specifically, we ... | Using the gradients of activation functions to back propagate relevance can sometimes lead to misleading contribution attributions due to discontinuous or vanishing gradients. In order to overcome this, instead of using the gradients to back propagate the relevance, Deep Learning Important FeaTures (DeepLIFT) @cite_21 ... | 891 | 144 |
Convolutional Neural Networks (CNN) have become state-of-the-art in the field of image classification. However, not everything is understood about their inner representations. This paper tackles the interpretability and explainability of the predictions of CNNs for multi-class classification problems. Specifically, we ... | Gradient-Weighted CAM (Grad-CAM) @cite_24 is a generalization of CAM that can target any layer and introduces the gradient information to CAM. The problem with CAM-based methods is that they specifically target high-level layers. Therefore, hybrid methods, such as Guided Grad-CAM @cite_24 , combine the qualities of CAM... | 728 | 100 |
Convolutional Neural Networks (CNN) have become state-of-the-art in the field of image classification. However, not everything is understood about their inner representations. This paper tackles the interpretability and explainability of the predictions of CNNs for multi-class classification problems. Specifically, we ... | There are also many other visualization and explanation methods for neural networks. For example, by observing the maximal activations of layers, it is possible to visualize contributing regions of feature maps @cite_36 @cite_7 or the use of attention visualization @cite_30 @cite_34 . In addition, the hidden layers of ... | 1,450 | 132 |
Predicting the future of Graph-supported Time Series (GTS) is a key challenge in many domains, such as climate monitoring, finance or neuroimaging. Yet it is a highly difficult problem as it requires to account jointly for time and graph (spatial) dependencies. To simplify this process, it is common to use a two-step p... | When it comes to graphs, a natural approach to define a latent representation is a structure-based approach. It consists in defining the notion of frequency on a graph by analogy with the Fourier Transform (FT) @cite_2 . The so-called Graph Fourier Transform (GFT) represents the signal in another basis. This basis is b... | 451 | 255 |
Predicting the future of Graph-supported Time Series (GTS) is a key challenge in many domains, such as climate monitoring, finance or neuroimaging. Yet it is a highly difficult problem as it requires to account jointly for time and graph (spatial) dependencies. To simplify this process, it is common to use a two-step p... | Finding the best graph structure for a given problem is a question that recently sparked a lot of interest in the literature @cite_7 @cite_11 @cite_12 . In many cases, a good solution consists in using the covariance matrix (or its inverse) as the adjacency matrix of the graph. Following this lead, in this paper, we ma... | 718 | 174 |
Predicting the future of Graph-supported Time Series (GTS) is a key challenge in many domains, such as climate monitoring, finance or neuroimaging. Yet it is a highly difficult problem as it requires to account jointly for time and graph (spatial) dependencies. To simplify this process, it is common to use a two-step p... | In this work, much of the interest in the latent representation is motivated by its use for sequence time prediction. There are, again, several recent works in this area, based on diverse methods (dictionary learning @cite_4 , source separation @cite_3 ). Again, we restrict the study to a simple question: what is the b... | 624 | 98 |
When people browse online news, small thumbnail images accompanying links to articles attract their attention and help them to decide which articles to read. As an increasing proportion of online news can be construed as data journalism, we have witnessed a corresponding increase in the incorporation of visualization i... | Prior research shows that the presence of thumbnails in search results help people locate articles of interest online @cite_19 @cite_21 @cite_8 , particularly when paired with informative titles, text snippets, and URLs. Thumbnails are a particularly important signal of relevance when some links to articles have them a... | 1,096 | 162 |
When people browse online news, small thumbnail images accompanying links to articles attract their attention and help them to decide which articles to read. As an increasing proportion of online news can be construed as data journalism, we have witnessed a corresponding increase in the incorporation of visualization i... | Beyond web search and email reading, thumbnails also appear in the context of navigating other forms of media, from file systems @cite_9 to documents @cite_13 and videos @cite_0 . Prior work in the visualization and visual analytics community has also incorporated thumbnails into the sensemaking process as a means of l... | 1,151 | 91 |
When people browse online news, small thumbnail images accompanying links to articles attract their attention and help them to decide which articles to read. As an increasing proportion of online news can be construed as data journalism, we have witnessed a corresponding increase in the incorporation of visualization i... | When browsing unfamiliar content, such as in the case of online news reading, thumbnails must compete for the reader's attention with one another and with other content. We therefore turn to other prior research examining specific aspects of thumbnail design. Several factors appear to impact how thumbnails draw attenti... | 353 | 142 |
When people browse online news, small thumbnail images accompanying links to articles attract their attention and help them to decide which articles to read. As an increasing proportion of online news can be construed as data journalism, we have witnessed a corresponding increase in the incorporation of visualization i... | Other prior research has considered how to automatically generate thumbnails using photos and images appearing in the article, such as by cropping, resizing, or selecting the most salient excerpts from them @cite_5 @cite_30 . More recently, @cite_37 introduced an algorithm to select highly salient and evocative thumbna... | 875 | 117 |
When people browse online news, small thumbnail images accompanying links to articles attract their attention and help them to decide which articles to read. As an increasing proportion of online news can be construed as data journalism, we have witnessed a corresponding increase in the incorporation of visualization i... | Visualization is increasingly prevalent in news media @cite_36 . In this context, the communicative intent of visualization often leads to different design choices than those used in the context of data analysis @cite_33 . As a result, we encounter substantial use of graphical and text-based annotation @cite_26 . We al... | 2,440 | 221 |
A generalized @math puzzle' consists of an @math numbered grid, with one missing number. A move in the game switches the position of the empty square with the position of one of its neighbors. We solve Diaconis' 15 puzzle problem' by proving that the asymptotic total variation mixing time of the board is at least order... | Much of the prior work regarding @math puzzles has focused on sorting strategies for a puzzle in general position, and on the hardness of finding shortest sorting algorithms. For instance, it is known that any @math puzzle may be returned to sorted order in order @math steps (and fewer are not always possible) @cite_14... | 498 | 103 |
A generalized @math puzzle' consists of an @math numbered grid, with one missing number. A move in the game switches the position of the empty square with the position of one of its neighbors. We solve Diaconis' 15 puzzle problem' by proving that the asymptotic total variation mixing time of the board is at least order... | A host of random walks on permutation groups have been studied via a variety of techniques, including the representation theory @cite_8 , @cite_15 , @cite_10 , and couplings @cite_11 . Our method for the upper bound, which uses a three transitive group action, is based on @cite_9 . So far as we are aware, the method of... | 1,150 | 135 |
We present a learning-based method to estimate the object bounding box from its 2D bird's-eye view (BEV) LiDAR points. Our method, entitled BoxNet, exploits a simple deep neural network that can efficiently handle unordered points. The method takes as input the 2D coordinates of all the points and the output is a vecto... | We omit the extensive literature review of 3D object detection methods, in which bounding box regression is considered as an essential component. Our proposed method only focuses on the 2D case and does not target an end-to-end solution. Therefore, it is in fact an intermediate step in the pipeline of 3D object detecti... | 482 | 139 |
Abstract In two earlier papers we derived congruence formats with regard to transition system specifications for weak semantics on the basis of a decomposition method for modal formulas. The idea is that a congruence format for a semantics must ensure that the formulas in the modal characterisation of this semantics ar... | In @cite_30 @cite_0 he employs (OSOS) TSSs @cite_4 . An OSOS TSS allows no negative premises, but includes priorities between rules: @math means that @math can only be applied if @math cannot. An OSOS specification can be seen as, or translated into, a GSOS specification with negative premises. Each rule @math with exa... | 705 | 174 |
Abstract In two earlier papers we derived congruence formats with regard to transition system specifications for weak semantics on the basis of a decomposition method for modal formulas. The idea is that a congruence format for a semantics must ensure that the formulas in the modal characterisation of this semantics ar... | If patience rules are not allowed to have a lower priority than other rules, then the (r)bbo format, upon translation from OSOS to GSOS, can be seen as a subformat of our (rooted) stability-respecting branching bisimulation format. The basic idea is that in the rbbo format all arguments of so-called @math -preserving f... | 701 | 324 |
Until now, researchers have proposed several novel heterogeneous defect prediction HDP methods with promising performance. To the best of our knowledge, whether HDP methods can perform significantly better than unsupervised methods has not yet been thoroughly investigated. In this article, we perform a replication stud... | Researchers conducted a set of empirical studies to investigate the feasibility of CPDP by considering real-world software projects. @cite_75 analyzed 12 real-world projects from open-source communities and Microsoft corporation. After running 622 cross-project predictions, they found only 3.4 , @cite_65 analyzed anoth... | 2,175 | 260 |
In automatic speech recognition, often little training data is available for specific challenging tasks, but training of state-of-the-art automatic speech recognition systems requires large amounts of annotated speech. To address this issue, we propose a two-staged approach to acoustic modeling that combines noise and ... | Applying data augmentation to training data is a common approach to increase the amount of training data in order to improve the robustness of a model. In ASR it can be used, e.g., to apply multi-condition training, when no real data in the desired condition is available. Data augmentation is, however, limited to acous... | 471 | 213 |
In automatic speech recognition, often little training data is available for specific challenging tasks, but training of state-of-the-art automatic speech recognition systems requires large amounts of annotated speech. To address this issue, we propose a two-staged approach to acoustic modeling that combines noise and ... | Transfer learning is an approach used to transfer knowledge of a model trained in one scenario to train a model in another related scenario to improve generalization and performance @cite_14 . It is particularly useful in scenarios where only little training data is available for the main task but a large amount of ann... | 493 | 126 |
First, a new perspective based on binary matrices of placement delivery array (PDA) design was introduced, by which the PDA design problem can be simplified. From this new perspective, and based on some families of combinatorial designs, new schemes with low subpacketization for centralized coded caching problem were c... | Although optimal in rate, the caching scheme in @cite_22 has its limitation in practical implementations: By this caching scheme, each file is divided into @math packets @math The number @math is also referred to as the file size or subpacketization in some literature. @math , which grows exponentially with @math @cite... | 796 | 104 |
First, a new perspective based on binary matrices of placement delivery array (PDA) design was introduced, by which the PDA design problem can be simplified. From this new perspective, and based on some families of combinatorial designs, new schemes with low subpacketization for centralized coded caching problem were c... | So far, several coded caching schemes with reduced file size have been constructed, all with the sacrifice of increasing the rate. In @cite_2 , a class of coded caching schemes with linear file size @math i.e., @math were constructed from Ruzsa-Szem @math redi graphs. A very interesting framework for constructing centr... | 1,760 | 352 |
Energy minimization methods are a classical tool in a multitude of computer vision applications. While they are interpretable and well-studied, their regularity assumptions are difficult to design by hand. Deep learning techniques on the other hand are purely data-driven, often provide excellent results, but are very d... | The straightforward way of optimizing bi-level problems is to consider @cite_49 @cite_2 @cite_57 . These methods directly differentiate the higher-level loss function with respect to the minimizing argument and descend in the direction of this gradient. An incomplete list of examples in image processing is @cite_29 @ci... | 2,293 | 260 |
Energy minimization methods are a classical tool in a multitude of computer vision applications. While they are interpretable and well-studied, their regularity assumptions are difficult to design by hand. Deep learning techniques on the other hand are purely data-driven, often provide excellent results, but are very d... | is a prominent strategy in applied bi-level optimization across fields, i.e. MRF literature @cite_59 @cite_10 in deep learning @cite_11 @cite_28 @cite_14 @cite_87 and in variational settings @cite_71 @cite_97 @cite_54 @cite_26 @cite_75 @cite_39 . The problem is transformed into a single level problem by choosing an opt... | 3,526 | 286 |
Let @math be a discrete time Markov chain on a general state space. It is well-known that if @math is aperiodic and satisfies a drift and minorization condition, then it converges to its stationary distribution @math at an exponential rate. We consider the problem of computing upper bounds for the distance from station... | In the case @math , the decay rate @math of the law of @math was identified by Roberts and Tweedie @cite_18 (who use the notation @math ). Theorem 4.1(i) of @cite_18 is equivalent to a bound of the form [ (T > t) ( const ) (V)^r t ^t. ] Theorem slightly improves this result by removing the factor of @math and generaliz... | 523 | 124 |
Let @math be a discrete time Markov chain on a general state space. It is well-known that if @math is aperiodic and satisfies a drift and minorization condition, then it converges to its stationary distribution @math at an exponential rate. We consider the problem of computing upper bounds for the distance from station... | The most important feature of Theorem and its @math -norm version, Theorem , is that the exponential rate @math is the same as the decay rate in Theorem . As we will see in Section , this conclusion can only be drawn for reversible Markov chains with nonnegative eigenvalues and does not hold in general. Baxendale @cite... | 435 | 202 |
Let @math be a discrete time Markov chain on a general state space. It is well-known that if @math is aperiodic and satisfies a drift and minorization condition, then it converges to its stationary distribution @math at an exponential rate. We consider the problem of computing upper bounds for the distance from station... | The method of proof in @cite_11 uses analytic properties of generating functions for renewal sequences. In principle the argument could be extended to the case @math , but the resulting bound on the exponential convergence rate of the chain would be worse than the rate @math from Theorem . Intuitively, this is because ... | 435 | 112 |
Recent applications of unmanned aerial systems (UAS) to precision agriculture have shown increased ease and efficiency in data collection at precise remote locations. However, further enhancement of the field requires operation over long periods of time, e.g. days or weeks. This has so far been impractical due to the l... | Several rotorcraft UAS that are capable of autonomous, long-duration mission execution in benign indoor (VICON) environments have previously appeared in literature. Focusing on the recharging solution to extend individual platform flight time and a multi-agent scheme for constant operation, impressive operation times h... | 1,458 | 274 |
Recent applications of unmanned aerial systems (UAS) to precision agriculture have shown increased ease and efficiency in data collection at precise remote locations. However, further enhancement of the field requires operation over long periods of time, e.g. days or weeks. This has so far been impractical due to the l... | Our work uses a downward-facing monocular camera to estimate the pose of the landing pad in the world frame using AprilTag visual fiducial markers @cite_60 . We believe that a monocular camera is the cheapest, most lightweight and power efficient sensor choice. Alternatively, GPS and RTK-GPS systems suffer from precisi... | 731 | 134 |
Recent applications of unmanned aerial systems (UAS) to precision agriculture have shown increased ease and efficiency in data collection at precise remote locations. However, further enhancement of the field requires operation over long periods of time, e.g. days or weeks. This has so far been impractical due to the l... | Several landing approaches exist in literature for labeled and unlabeled landing sites. @cite_34 present a monocular visual landing method based on estimating the 6 DOF pose of a circled H marker. The same authors extend this work to enable autonomous landing site search by using a scale-corrected PTAM algorithm @cite_... | 1,953 | 236 |
Recent applications of unmanned aerial systems (UAS) to precision agriculture have shown increased ease and efficiency in data collection at precise remote locations. However, further enhancement of the field requires operation over long periods of time, e.g. days or weeks. This has so far been impractical due to the l... | Currently one of the most popular visual fiducial detectors and patterns is the AprilTag algorithm @cite_15 which is renowned for its speed, robustness and extremely low false positive detection rates. The algorithm was updated by @cite_60 to further improve computational efficiency and to enable the detection of small... | 1,250 | 252 |
Recent applications of unmanned aerial systems (UAS) to precision agriculture have shown increased ease and efficiency in data collection at precise remote locations. However, further enhancement of the field requires operation over long periods of time, e.g. days or weeks. This has so far been impractical due to the l... | After obtaining raw tag pose estimates from the AprilTag detector, our approach uses a recursive least squares (RLS) filter to obtain a common tag bundle pose estimate. A relevant previous work on this topic is @cite_55 in which a particle filter is applied to raw tag detections and RLS is compared to RANSAC for bundle... | 746 | 223 |
Recent applications of unmanned aerial systems (UAS) to precision agriculture have shown increased ease and efficiency in data collection at precise remote locations. However, further enhancement of the field requires operation over long periods of time, e.g. days or weeks. This has so far been impractical due to the l... | Our approach, however, is not limited to AprilTags, which can be substituted or combined with other markers for specific applications. A vast number of markers is available targeting different use-cases @cite_42 @cite_34 @cite_12 @cite_15 . Patterns include letters, circles, concentric circles and or polygons, letters,... | 1,786 | 122 |
Most graph neural networks can be described in terms of message passing, vertex update, and readout functions. In this paper, we represent documents as word co-occurrence networks and propose an application of the message passing framework to NLP, the Message Passing Attention network for Document understanding (MPAD).... | There are significant differences between @cite_51 and our work. First, our approach is Note that other GNNs used in inductive settings can be found @cite_10 @cite_52 . , not . Indeed, while the node classification approach of requires all test documents at training time, our graph classification model is able to perfo... | 902 | 181 |
External knowledge is often useful for natural language understanding tasks. We introduce a contextual text representation model called Conceptual-Contextual (CC) embeddings, which incorporates structured knowledge into text representations. Unlike entity embedding methods, our approach encodes a knowledge graph into a... | In recent years a number of KB embedding models have been proposed that aim at learning entity embeddings on a knowledge graph @cite_18 . Some models make use of textual information in KBs to improve entity embeddings, like using textual descriptions of entities as complement to triplet modeling @cite_30 @cite_5 , or j... | 1,016 | 177 |
External knowledge is often useful for natural language understanding tasks. We introduce a contextual text representation model called Conceptual-Contextual (CC) embeddings, which incorporates structured knowledge into text representations. Unlike entity embedding methods, our approach encodes a knowledge graph into a... | Recently contextual text representation models like ELMo , BERT @cite_10 and OpenAI GPT @cite_8 @cite_28 have pushed the state-of-the-art results of various NLP tasks. Language modeling on a giant corpus learns powerful representations, which provides huge benefits to supervised tasks, especially where labeled data is ... | 820 | 153 |
Graph drawing and visualisation techniques are important tools for the exploratory analysis of complex systems. While these methods are regularly applied to visualise data on complex networks, we increasingly have access to time series data that can be modelled as temporal networks or dynamic graphs. In such dynamic gr... | Having motivated the effects that are due to the arrow of time in dynamic graphs, we review related works on dynamic graph drawing. Using the taxonomy from @cite_35 , we categorize those works in (i) animation techniques that map the time dimension of dynamic graphs onto a time dimension of the resulting visualisation,... | 617 | 138 |
Graph drawing and visualisation techniques are important tools for the exploratory analysis of complex systems. While these methods are regularly applied to visualise data on complex networks, we increasingly have access to time series data that can be modelled as temporal networks or dynamic graphs. In such dynamic gr... | Apart from the issue that animations are cognitively demanding, additional challenges arise in data with high temporal resolution (e.g. seconds or even millisecond), where a single vertex or edge is likely to be active in each time stamp. The application of static graph drawing techniques to such data requires a coarse... | 379 | 150 |
Graph drawing and visualisation techniques are important tools for the exploratory analysis of complex systems. While these methods are regularly applied to visualise data on complex networks, we increasingly have access to time series data that can be modelled as temporal networks or dynamic graphs. In such dynamic gr... | Despite the advances outlined above, recognizing patterns in animation-based visualisations of dynamic graphs remains a considerable cognitive challenge for users. Moreover, it is difficult to embed dynamic graph animations into scholarly articles, books, or posters, which often limits their use in science and engineer... | 2,066 | 296 |
Style is ubiquitous in our daily language uses, while what is language style to learning machines? In this paper, by exploiting the second-order statistics of semantic vectors of different corpora, we present a novel perspective on this question via style matrix, i.e. the covariance matrix of semantic vectors, and expl... | Similarly, image style transfer aims to reconstruct an image with some characters of the style image while preserving its content. The groundbreaking works proposed by @cite_0 @cite_18 show that the Gram matrices (or covariance matrices) of the feature maps, which are extracted by a frozen convolution neural network tr... | 1,354 | 285 |
3D Human Motion Indexing and Retrieval is an interesting problem due to the rise of several data-driven applications aimed at analyzing and or re-utilizing 3D human skelet al data, such as data-driven animation, analysis of sports bio-mechanics, human surveillance etc. Spatio-temporal articulations of humans, noisy mis... | Most approaches on the 3D human motion retrieval have focused on developing hand crafted features to represent the skeleton sequences @cite_17 @cite_19 @cite_4 . In this section, we broadly categorize them by the method in which they engineer their descriptors. | 1,122 | 64 |
3D Human Motion Indexing and Retrieval is an interesting problem due to the rise of several data-driven applications aimed at analyzing and or re-utilizing 3D human skelet al data, such as data-driven animation, analysis of sports bio-mechanics, human surveillance etc. Spatio-temporal articulations of humans, noisy mis... | For the task of retrieval, @cite_2 proposed a simple auto-encoder that captures high-level features. However, their model doesn't explicitly use a temporal construct for motion data. Primarily, learnable representations from 3D motion data have been used for other tasks. @cite_18 @cite_14 are a few amongst many who use... | 1,167 | 143 |
We study the fair allocation of indivisible goods under the assumption that the goods form an undirected graph and each agent must receive a connected subgraph. Our focus is on well-studied fairness notions including envy-freeness and maximin share fairness. We establish graph-specific maximin share guarantees, which a... | Fair allocation of indivisible goods has received considerable attention from the research community, especially in the last few years. We refer to surveys by @cite_7 , @cite_10 , and @cite_14 for an overview of recent developments in the area. | 405 | 56 |
We study the fair allocation of indivisible goods under the assumption that the goods form an undirected graph and each agent must receive a connected subgraph. Our focus is on well-studied fairness notions including envy-freeness and maximin share fairness. We establish graph-specific maximin share guarantees, which a... | The papers most closely related to ours are the two papers that we mentioned, by @cite_12 and @cite_5 . showed that for any number of agents with additive valuations, there always exists an allocation that gives every agent her maximin share when the graph is a tree, but not necessarily when the graph is a cycle. It is... | 959 | 187 |
We study the fair allocation of indivisible goods under the assumption that the goods form an undirected graph and each agent must receive a connected subgraph. Our focus is on well-studied fairness notions including envy-freeness and maximin share fairness. We establish graph-specific maximin share guarantees, which a... | @cite_5 investigated the same model with respect to relaxations of envy-freeness. As we mentioned, they characterized the set of graphs for which EF1 can be guaranteed in the case of two agents with arbitrary monotonic valuations. Moreover, they showed that an EF1 allocation always exists on a path for @math . Intrigui... | 610 | 112 |
We study the fair allocation of indivisible goods under the assumption that the goods form an undirected graph and each agent must receive a connected subgraph. Our focus is on well-studied fairness notions including envy-freeness and maximin share fairness. We establish graph-specific maximin share guarantees, which a... | Besides @cite_12 and @cite_5 , a number of other authors have recently studied fairness under connectivity constraints. @cite_4 investigated maximin share fairness in the case of cycles, also concentrating on the G-MMS notion, while @cite_9 focused on paths and provided approximations of envy-freeness, proportionality,... | 1,787 | 160 |
We study the fair allocation of indivisible goods under the assumption that the goods form an undirected graph and each agent must receive a connected subgraph. Our focus is on well-studied fairness notions including envy-freeness and maximin share fairness. We establish graph-specific maximin share guarantees, which a... | A related line of work also combines graphs with resource allocation, but uses graphs to capture the connection between agents instead of goods. In particular, a graph specifies the acquaintance relationship among agents. @cite_1 and @cite_2 defined graph-based versions of envy-freeness and proportionality with divisib... | 1,174 | 131 |
Musculoskelet al robots that are based on pneumatic actuation have a variety of properties, such as compliance and back-drivability, that render them particularly appealing for human-robot collaboration. However, programming interactive and responsive behaviors for such systems is extremely challenging due to the nonli... | Robots with pneumatic artificial muscles (PAMs) and compliant limbs have been shown to be desirable for human-robot interaction scenarios @cite_22 @cite_13 . When configured in an anthropomorphic musculoskelet al structure, such robots provide an intriguing platform for human-robot interaction (HRI) @cite_4 due to thei... | 1,077 | 279 |
Prior work has demonstrated that question classification (QC), recognizing the problem domain of a question, can help answer it more accurately. However, developing strong QC algorithms has been hindered by the limited size and complexity of annotated data available. To address this, we present the largest challenge da... | Question classification typically makes use of a combination of syntactic, semantic, surface, and embedding methods. Syntactic patterns @cite_25 @cite_37 @cite_28 @cite_7 and syntactic dependencies @cite_2 have been shown to improve performance, while syntactically or semantically important words are often expanding us... | 2,303 | 226 |
Real-time 3D reconstruction from RGB-D sensor data plays an important role in many robotic applications, such as object modeling and mapping. The popular method of fusing depth information into a truncated signed distance function (TSDF) and applying the marching cubes algorithm for mesh extraction has severe issues wi... | Surface reconstruction from range data has been an active research topic for a long time. It gained in popularity through the availability of affordable depth cameras and parallel computing hardware. Zollhöfer al @cite_10 give a comprehensive overview on modern 3D reconstruction from RGB-D data. The two main streams or... | 1,195 | 170 |
Real-time 3D reconstruction from RGB-D sensor data plays an important role in many robotic applications, such as object modeling and mapping. The popular method of fusing depth information into a truncated signed distance function (TSDF) and applying the marching cubes algorithm for mesh extraction has severe issues wi... | In contrast to the related works, we are proposing an improved representation based on the TSDF that utilizes the idea from Henry @cite_12 to represent surfaces with different orientations separate from each other. The implementation is based on the work of Dong al @cite_1 , which also serves as a baseline for state-of... | 655 | 194 |
We investigate the performance of a jet identification algorithm based on interaction networks (JEDI-net) to identify all-hadronic decays of high-momentum heavy particles produced at the LHC and distinguish them from ordinary jets originating from the hadronization of quarks and gluons. The jet dynamics is described as... | Jet tagging is one of the most popular LHC-related tasks to which DL solutions have been applied. Several classification algorithms have been studied in the context of jet tagging at the LHC @cite_47 @cite_1 @cite_5 @cite_10 @cite_2 @cite_12 @cite_43 @cite_13 using DNNs, CNNs, or physics-inspired architectures. Recurre... | 2,723 | 279 |
Recent advances in deep learning greatly boost the performance of object detection. State-of-the-art methods such as Faster-RCNN, FPN and R-FCN have achieved high accuracy in challenging benchmark datasets. However, these methods require fully annotated object bounding boxes for training, which are incredibly hard to s... | There are only a few research works in the semi-supervised detection field. @cite_35 proposes a LSDA-based method that can handle disjoint set semi-supervised detection, but this method is not end-to-end trainable and cannot be easily extended to state-of-the-art detection frameworks. @cite_20 proposes a semiMIL method... | 1,499 | 195 |
The spectacular success of Bitcoin and Blockchain Technology in recent years has provided enough evidence that a widespread adoption of a common cryptocurrency system is not merely a distant vision, but a scenario that might come true in the near future. However, the presence of Bitcoin's obvious shortcomings such as e... | Atomic Broadcast. For an excellent introduction to the field of Distributed Computing and overview of Atomic Broadcast and Consensus protocols we refer the reader to the book @cite_53 . A more recent work of @cite_6 surveys existing consensus protocols in the context of cryptocurrency systems. | 788 | 63 |
The spectacular success of Bitcoin and Blockchain Technology in recent years has provided enough evidence that a widespread adoption of a common cryptocurrency system is not merely a distant vision, but a scenario that might come true in the near future. However, the presence of Bitcoin's obvious shortcomings such as e... | In this paper we propose a different assumption on the transaction buffers that allows us to better demonstrate the capabilities of our protocol when it comes to latency. We assume that at every round the ratio between lengths of transaction buffers of any two honest nodes is at most a fixed constant. In this model, ou... | 555 | 197 |
Re-Pair is a grammar compression scheme with favorably good compression rates. The computation of Re-Pair comes with the cost of maintaining large frequency tables, which makes it hard to compute Re-Pair on large scale data sets. As a solution for this problem we present, given a text of length @math whose characters a... | In-Place String Algorithms For the LZ77 factorization @cite_7 , present an algorithm computing this factorization with n d words on top of the input space in dn time for a variable @math , achieving 1 words with n^2 time. For the suffix sorting problem, gave an algorithm to compute the suffix array with n bits on top o... | 349 | 179 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.