aid stringlengths 9 15 | mid stringlengths 7 10 | abstract stringlengths 78 2.56k | related_work stringlengths 92 1.77k | ref_abstract dict |
|---|---|---|---|---|
1907.09189 | 2963689651 | This paper is concerned with evaluating different multiagent learning (MAL) algorithms in problems where individual agents may be heterogenous, in the sense of utilizing different learning strategies, without the opportunity for prior agreements or information regarding coordination. Such a situation arises in ad hoc t... | The problem of incomplete information in multiagent learning, in the form of the ad hoc team problem, was addressed by @cite_20 . They propose a procedure to evaluate two ad hoc agents for a given set of potential team members and tasks. We used a modified version of this procedure for our own experiments (see ). | {
"cite_N": [
"@cite_20"
],
"mid": [
"1606056663"
],
"abstract": [
"As autonomous agents proliferate in the real world, both in software and robotic settings, they will increasingly need to band together for cooperative activities with previously unfamiliar teammates. In such ad hoc team settings,... |
1907.09189 | 2963689651 | This paper is concerned with evaluating different multiagent learning (MAL) algorithms in problems where individual agents may be heterogenous, in the sense of utilizing different learning strategies, without the opportunity for prior agreements or information regarding coordination. Such a situation arises in ad hoc t... | In earlier work, Stone and Kraus @cite_14 define optimal strategies for an ad hoc agent collaborating with a fixed-behaviour teammate in an environment modelled as a @math -armed bandit. @cite_32 present an algorithm that would lead a fixed greedy agent towards an optimal joint action in a simple repeated game in which... | {
"cite_N": [
"@cite_14",
"@cite_32"
],
"mid": [
"1548012162",
"2141088582"
],
"abstract": [
"In typical multiagent teamwork settings, the teammates are either programmed together, or are otherwise provided with standard communication languages and coordination protocols. In contrast, this... |
1907.09189 | 2963689651 | This paper is concerned with evaluating different multiagent learning (MAL) algorithms in problems where individual agents may be heterogenous, in the sense of utilizing different learning strategies, without the opportunity for prior agreements or information regarding coordination. Such a situation arises in ad hoc t... | These assumptions are relaxed in a recent empirical study by @cite_28 . They used an ad hoc agent that tries to identify its teammates by observing their behaviour and comparing it with a database of known behaviours. In addition, it learns a new model for the observed behaviour using a tree classifier. The agent combi... | {
"cite_N": [
"@cite_28"
],
"mid": [
"2139993574"
],
"abstract": [
"The concept of creating autonomous agents capable of exhibiting ad hoc teamwork was recently introduced as a challenge to the AI, and specifically to the multiagent systems community. An agent capable of ad hoc teamwork is one tha... |
1907.09189 | 2963689651 | This paper is concerned with evaluating different multiagent learning (MAL) algorithms in problems where individual agents may be heterogenous, in the sense of utilizing different learning strategies, without the opportunity for prior agreements or information regarding coordination. Such a situation arises in ad hoc t... | @cite_13 proposed an interesting algorithm called (OPAT). For each encountered state, the algorithm estimates the values of all joint actions using Monte-Carlo Tree Search. These values are used to generate a stage game (i.e. a repeated game with one repetition), based on which the algorithm decides which action to tak... | {
"cite_N": [
"@cite_13"
],
"mid": [
"160961942"
],
"abstract": [
"We propose a novel online planning algorithm for ad hoc team settings--challenging situations in which an agent must collaborate with unknown teammates without prior coordination. Our approach is based on constructing and solving a... |
1907.09211 | 2963774932 | Network slicing appears as a key enabler for the future 5G networks. Mobile Network Operators create various slices for Service Providers (SP) to accommodate customized services. As network slices are operated on a common network infrastructure owned by some Infrastructure Provider (InP), sharing the resources across a... | Early results on assigning infrastructure network resources to virtual network components may be found, , in @cite_22 @cite_17 . Due to its capability of sharing efficiently network resource in 5G networks, the concept of network virtualization has gained renewed attention in the literature @cite_13 @cite_15 @cite_21 @... | {
"cite_N": [
"@cite_22",
"@cite_21",
"@cite_39",
"@cite_15",
"@cite_13",
"@cite_17"
],
"mid": [
"2142547489",
"2588061367",
"2793519446",
"2612074600",
"2605961225",
""
],
"abstract": [
"Recent proposals for network virtualization provide a promising way to... |
1907.09211 | 2963774932 | Network slicing appears as a key enabler for the future 5G networks. Mobile Network Operators create various slices for Service Providers (SP) to accommodate customized services. As network slices are operated on a common network infrastructure owned by some Infrastructure Provider (InP), sharing the resources across a... | Network slice resource allocation is a complex problem. When a slice instance is seen as a collection of SFCs, slice embedding needs to deploy the SFCs on a shared infrastructure while satisfying various constraints. Most of prior works related to SFC and VNF deployment do not account for coverage constraints. For exam... | {
"cite_N": [
"@cite_14",
"@cite_28",
"@cite_3",
"@cite_0",
"@cite_24",
"@cite_27",
"@cite_2",
"@cite_17"
],
"mid": [
"2334600287",
"2964125711",
"2474498303",
"2739867585",
"2962948535",
"2132238781",
"1578960134",
""
],
"abstract": [
"Netwo... |
1907.09211 | 2963774932 | Network slicing appears as a key enabler for the future 5G networks. Mobile Network Operators create various slices for Service Providers (SP) to accommodate customized services. As network slices are operated on a common network infrastructure owned by some Infrastructure Provider (InP), sharing the resources across a... | To address the high computational complexity resulting from the ILPs or MILPs, various heuristics have been proposed, see, , @cite_14 @cite_0 @cite_2 . For example, @cite_14 introduced an heuristic based on the search of shortest paths to sequentially embed the SFCs. In @cite_0 , the candidate infrastructure nodes are ... | {
"cite_N": [
"@cite_0",
"@cite_14",
"@cite_2"
],
"mid": [
"2739867585",
"2334600287",
"1578960134"
],
"abstract": [
"With Network Function Virtualization (NFV), network functions are deployed as modular software components on the commodity hardware, and can be further chained to p... |
1907.09211 | 2963774932 | Network slicing appears as a key enabler for the future 5G networks. Mobile Network Operators create various slices for Service Providers (SP) to accommodate customized services. As network slices are operated on a common network infrastructure owned by some Infrastructure Provider (InP), sharing the resources across a... | In @cite_35 , the join VNF and virtual link placement is formulated as a (WGMP), where the SFC graph and the infrastructure graph are modeled as weighted graphs, on which each node and each link have their own weight corresponding to their required resource (for the SFC graph), or their available resource (for the infr... | {
"cite_N": [
"@cite_35",
"@cite_10",
"@cite_23"
],
"mid": [
"2490169957",
"2725479901",
"2623646697"
],
"abstract": [
"Network function virtualization (NFV) decouples software implementations of network functions from their hosts (or hardware). NFV exposes a new set of entities, t... |
1907.09211 | 2963774932 | Network slicing appears as a key enabler for the future 5G networks. Mobile Network Operators create various slices for Service Providers (SP) to accommodate customized services. As network slices are operated on a common network infrastructure owned by some Infrastructure Provider (InP), sharing the resources across a... | The design of efficient allocation mechanisms for virtualized radio resources has been recently addressed in @cite_4 . This paper aims at minimizing the leasing cost of BSs so as to meet SP demands, while providing, with a given probability, a minimum data rate for any user located in their coverage area. The rate cons... | {
"cite_N": [
"@cite_4"
],
"mid": [
"2963806969"
],
"abstract": [
"Wireless network virtualization is emerging as an important technology for next-generation (5G) wireless networks. A key advantage of introducing virtualization in cellular networks is that service providers can robustly share virt... |
1907.09211 | 2963774932 | Network slicing appears as a key enabler for the future 5G networks. Mobile Network Operators create various slices for Service Providers (SP) to accommodate customized services. As network slices are operated on a common network infrastructure owned by some Infrastructure Provider (InP), sharing the resources across a... | In @cite_36 , an heterogeneous spatial user density is considered, and the joint BS selection and adaptive slicing are formulated as a two-stage stochastic optimization problem. The first stage aims at defining the set of BSs to activate. The second stage aims at allocating wireless resources of the BSs to each point o... | {
"cite_N": [
"@cite_36"
],
"mid": [
"2937641582"
],
"abstract": [
"Wireless network virtualization is a promising avenue of research for next-generation 5G cellular networks. Virtualization focuses on the concept of active resource sharing and the building of a network designed for specific deman... |
1907.09211 | 2963774932 | Network slicing appears as a key enabler for the future 5G networks. Mobile Network Operators create various slices for Service Providers (SP) to accommodate customized services. As network slices are operated on a common network infrastructure owned by some Infrastructure Provider (InP), sharing the resources across a... | In @cite_34 , a network slicing framework for (H-CRAN) is introduced. The sharing of radio resources in terms of data rate is considered, with some constraints related to the fronthaul capacity, the transmission power budget of RRHs, or the tolerable interference threshold of an RRH on a sub-channel. Slicing is formula... | {
"cite_N": [
"@cite_34"
],
"mid": [
"2601022114"
],
"abstract": [
"Research on network slicing for multi-tenant heterogeneous cloud radio access networks (H-CRANs) is still in its infancy. In this paper, we redefine network slicing and propose a new network slicing framework for multi-tenant H-CR... |
1907.09211 | 2963774932 | Network slicing appears as a key enabler for the future 5G networks. Mobile Network Operators create various slices for Service Providers (SP) to accommodate customized services. As network slices are operated on a common network infrastructure owned by some Infrastructure Provider (InP), sharing the resources across a... | The wireless network slicing problem is also addressed in @cite_26 . A game theory-based distributed algorithm to solve the problem is proposed. The proposed algorithm accounts for the limited availability of wireless resources and considers different aspects such as congestion, deployment costs and the RRH-user distan... | {
"cite_N": [
"@cite_26"
],
"mid": [
"2964112281"
],
"abstract": [
"Radio access network (RAN) slicing is an effective methodology to dynamically allocate networking resources in 5G networks. One of the main challenges of RAN slicing is that it is provably an NP-Hard problem. For this reason, we d... |
1907.09211 | 2963774932 | Network slicing appears as a key enabler for the future 5G networks. Mobile Network Operators create various slices for Service Providers (SP) to accommodate customized services. As network slices are operated on a common network infrastructure owned by some Infrastructure Provider (InP), sharing the resources across a... | Compared to previous works, we consider slice resource demands in terms of coverage and traffic requirements in the radio access part of the network as well as network, storage, and computing requirements from a cloud infrastructure of interconnected data centers for the rest of the network. This work borrows the slice... | {
"cite_N": [
"@cite_14",
"@cite_4",
"@cite_26",
"@cite_36",
"@cite_32",
"@cite_0",
"@cite_34",
"@cite_20"
],
"mid": [
"2334600287",
"2963806969",
"2964112281",
"2937641582",
"2913775178",
"2739867585",
"2601022114",
"2916661620"
],
"abstract": [... |
1907.09211 | 2963774932 | Network slicing appears as a key enabler for the future 5G networks. Mobile Network Operators create various slices for Service Providers (SP) to accommodate customized services. As network slices are operated on a common network infrastructure owned by some Infrastructure Provider (InP), sharing the resources across a... | In this work, we assume that the resource requirements for the various SFCs that will have to be deployed within a slice may be aggregated and represented by a Slice Resource Demand (SRD) graph that mimics the graph of SFCs. These SRDs are evaluated by the MNO to satisfy the QoS requirements imposed by the tenant. The ... | {
"cite_N": [
"@cite_0",
"@cite_14"
],
"mid": [
"2739867585",
"2334600287"
],
"abstract": [
"With Network Function Virtualization (NFV), network functions are deployed as modular software components on the commodity hardware, and can be further chained to provide services, offering much gr... |
1907.09211 | 2963774932 | Network slicing appears as a key enabler for the future 5G networks. Mobile Network Operators create various slices for Service Providers (SP) to accommodate customized services. As network slices are operated on a common network infrastructure owned by some Infrastructure Provider (InP), sharing the resources across a... | When provisioning slices, we consider coverage constraints, in which slices are assumed to cover a specific region in the considered geographical area, that is part of the SLA with the tenant. We devise the special case of the cloud RAN architecture with RRHs which are nodes having radio resources. In our model, radio ... | {
"cite_N": [
"@cite_4"
],
"mid": [
"2963806969"
],
"abstract": [
"Wireless network virtualization is emerging as an important technology for next-generation (5G) wireless networks. A key advantage of introducing virtualization in cellular networks is that service providers can robustly share virt... |
1907.09128 | 2963975574 | In this work, we present a modified fuzzy decision forest for real-time 3D object pose estimation based on typical template representation. We employ an extra preemptive background rejector node in the decision forest framework to terminate the examination of background locations as early as possible, result in a signi... | To effectively measure the similarity between object views, a compact and discriminative description vector is required. @cite_12 present a novel image representation, a rigid template using colour gradient and surface normal as feature descriptors called LineMOD. The templates are synthetically rendered from 3D object... | {
"cite_N": [
"@cite_12"
],
"mid": [
"1526868886"
],
"abstract": [
"We propose a framework for automatic modeling, detection, and tracking of 3D objects with a Kinect. The detection part is mainly based on the recent template-based LINEMOD approach [1] for object detection. We show how to build th... |
1907.09128 | 2963975574 | In this work, we present a modified fuzzy decision forest for real-time 3D object pose estimation based on typical template representation. We employ an extra preemptive background rejector node in the decision forest framework to terminate the examination of background locations as early as possible, result in a signi... | Hash table is a well-known data structure that allows a symbol lookup in @math complexity. In other words, the searching time is constant regardless of the database size. However, hash table is only able to find the exact match while in ANN searching problem we seek approximate matches. The most straight forward soluti... | {
"cite_N": [
"@cite_4"
],
"mid": [
"2963203908"
],
"abstract": [
"We present a scalable method for detecting objects and estimating their 3D poses in RGB-D data. To this end, we rely on an efficient representation of object views and employ hashing techniques to match these views against the inpu... |
1907.09128 | 2963975574 | In this work, we present a modified fuzzy decision forest for real-time 3D object pose estimation based on typical template representation. We employ an extra preemptive background rejector node in the decision forest framework to terminate the examination of background locations as early as possible, result in a signi... | The methods fall into this category focus on better generalisation to slight variations in translation, local shape and viewpoint. The explicit background foreground separation is learnt parametrically to deal with heavy background clutter. The result shows these approaches cause less false positives than nearest neigh... | {
"cite_N": [
"@cite_13",
"@cite_7"
],
"mid": [
"2050966058",
"1022526533"
],
"abstract": [
"In this paper we propose a new method for detecting multiple specific 3D objects in real time. We start from the template-based approach based on the LINE2D LINEMOD representation introduced recent... |
1907.09128 | 2963975574 | In this work, we present a modified fuzzy decision forest for real-time 3D object pose estimation based on typical template representation. We employ an extra preemptive background rejector node in the decision forest framework to terminate the examination of background locations as early as possible, result in a signi... | Registration-based methods attempt to fit a pose hypothesis to the observation, by iteratively update and minimise the discrepancy between the query sample and a sample rendered from the current pose hypothesis. A popular choice is the Iterative Closest Point (ICP) @cite_1 . | {
"cite_N": [
"@cite_1"
],
"mid": [
"2004312117"
],
"abstract": [
"Abstract This paper introduces a new method of registering point sets. The registration error is directly minimized using general-purpose non-linear optimization (the Levenberg–Marquardt algorithm). The surprising conclusion of the... |
1907.09128 | 2963975574 | In this work, we present a modified fuzzy decision forest for real-time 3D object pose estimation based on typical template representation. We employ an extra preemptive background rejector node in the decision forest framework to terminate the examination of background locations as early as possible, result in a signi... | One feasible solution is to cluster the templates into few sets, which has been proposed in few recent works. Hashmod @cite_4 clusters the templates with randomised forest and employs hashing techniques; Discriminatively Trained Templates (DTT) @cite_13 clusters the templates with a bottom-up clustering method and cons... | {
"cite_N": [
"@cite_13",
"@cite_4"
],
"mid": [
"2050966058",
"2963203908"
],
"abstract": [
"In this paper we propose a new method for detecting multiple specific 3D objects in real time. We start from the template-based approach based on the LINE2D LINEMOD representation introduced recent... |
1907.09081 | 2963906598 | Detecting objects in a two-dimensional setting is often insufficient in the context of real-life applications where the surrounding environment needs to be accurately recognized and oriented in three-dimension (3D), such as in the case of autonomous driving vehicles. Therefore, accurately and efficiently detecting obje... | The F-PointNet @cite_21 achieves notable performance for 3D object detection and birds-eye-view detection on cars, pedestrians and cyclists on the KITTI benchmark suite. This method uses a 2D Faster RCNN object detector to find 2D boxes including the object on RGB camera image. Subsequently, the detected boxes are extr... | {
"cite_N": [
"@cite_21"
],
"mid": [
"2769205412"
],
"abstract": [
"In this work, we study 3D object detection from RGB-D data in both indoor and outdoor scenes. While previous methods focus on images or 3D voxels, often obscuring natural 3D patterns and invariances of 3D data, we directly operate... |
1907.09150 | 2963415882 | We consider a generic empirical composition optimization problem, where there are empirical averages present both outside and inside nonlinear loss functions. Such a problem is of interest in various machine learning applications, and cannot be directly solved by standard methods such as stochastic gradient descent (SG... | Composition optimization have attracted significant attention in optimization literature. The stochastic version of the problem , where the empirical averages are replaced by expectations, is studied in @cite_19 . The authors propose a two-timescale stochastic approximation algorithm known as SCGD, and establish conver... | {
"cite_N": [
"@cite_14",
"@cite_4",
"@cite_1",
"@cite_19",
"@cite_11"
],
"mid": [
"2807364911",
"2765393021",
"2107438106",
"1494085563",
"2963008041"
],
"abstract": [
"We propose an accelerated stochastic compositional variance reduced gradient method for optimizi... |
1907.09150 | 2963415882 | We consider a generic empirical composition optimization problem, where there are empirical averages present both outside and inside nonlinear loss functions. Such a problem is of interest in various machine learning applications, and cannot be directly solved by standard methods such as stochastic gradient descent (SG... | Finally, our work is inspired by the stochastic variance reduction techniques in optimization @cite_0 @cite_1 @cite_23 @cite_3 @cite_6 , which considers the minimization of a cost that is a finite-sum of many component functions. Different versions of variance reduced stochastic gradients are constructed in these works... | {
"cite_N": [
"@cite_1",
"@cite_3",
"@cite_6",
"@cite_0",
"@cite_23"
],
"mid": [
"2107438106",
"2963357609",
"2758918273",
"2105875671",
"2135482703"
],
"abstract": [
"Stochastic gradient descent is popular for large scale optimization but has slow convergence asymp... |
1907.09369 | 2963317101 | In recent years, emotion detection in text has become more popular due to its vast potential applications in marketing, political science, psychology, human-computer interaction, artificial intelligence, etc. In this work, we argue that current methods which are based on conventional machine learning models cannot gras... | A lot of work has been done on detecting emotion in speech or visual data @cite_1 @cite_0 @cite_16 @cite_3 . But detecting emotions in textual data is a relatively new area that demands more research. There have been many attempts to detect emotions in text using conventional machine learning techniques and handcrafted... | {
"cite_N": [
"@cite_13",
"@cite_35",
"@cite_4",
"@cite_7",
"@cite_15",
"@cite_36",
"@cite_21",
"@cite_1",
"@cite_3",
"@cite_39",
"@cite_0",
"@cite_19",
"@cite_24",
"@cite_31",
"@cite_16",
"@cite_10",
"@cite_20"
],
"mid": [
"",
"",
"2... |
1907.09369 | 2963317101 | In recent years, emotion detection in text has become more popular due to its vast potential applications in marketing, political science, psychology, human-computer interaction, artificial intelligence, etc. In this work, we argue that current methods which are based on conventional machine learning models cannot gras... | Due to this sequential nature, recurrent and convolutional neural networks have been used in many NLP tasks and were able to improve the performance in a variety of classification tasks @cite_23 @cite_17 @cite_26 @cite_32 . There have been very few works in using deep neural network for emotion detection in text @cite_... | {
"cite_N": [
"@cite_26",
"@cite_29",
"@cite_32",
"@cite_23",
"@cite_2",
"@cite_17"
],
"mid": [
"2284289336",
"2606292552",
"2297405797",
"2265846598",
"2741447225",
""
],
"abstract": [
"Neural network models have been demonstrated to be capable of achieving... |
1907.09177 | 2963671871 | Advanced neural language models (NLMs) are widely used in sequence generation tasks because they are able to produce fluent and meaningful sentences. They can also be used to generate fake reviews, which can then be used to attack online review systems and influence the buying decisions of online shoppers. A problem in... | The most common attack on online review systems is a crowdturfing attack @cite_7 @cite_18 whereby a bad actor recruits a group of workers to write fake reviews based on a specified topic for a specified context and then submits them to the target website. Since this method has an economic cost, it is typically limited ... | {
"cite_N": [
"@cite_18",
"@cite_7"
],
"mid": [
"1916595307",
"2962991180"
],
"abstract": [
"Modern Web services inevitably engender abuse, as attackers find ways to exploit a service and its user base. However, while defending against such abuse is generally considered a technical endeavo... |
1907.09177 | 2963671871 | Advanced neural language models (NLMs) are widely used in sequence generation tasks because they are able to produce fluent and meaningful sentences. They can also be used to generate fake reviews, which can then be used to attack online review systems and influence the buying decisions of online shoppers. A problem in... | @cite_24 proposed such an attack method. Their idea is to first generate an initial fake review based on a given keyword using a long short-term memory (LSTM)-based LM. Because the initial fake review is stochastically sampled from a learned distribution, it may be irrelevant to the desired context. Then specific nouns... | {
"cite_N": [
"@cite_24",
"@cite_8"
],
"mid": [
"2752337926",
"2802987538"
],
"abstract": [
"Malicious crowdsourcing forums are gaining traction as sources of spreading misinformation online, but are limited by the costs of hiring and managing human workers. In this paper, we identify a ne... |
1907.09328 | 2962816295 | While search efficacy has been evaluated traditionally on the basis of result relevance, fairness of search has attracted recent attention. In this work, we define a notion of distributional fairness and provide a conceptual framework for evaluating search results based on it. As part of this, we formulate a set of axi... | Researchers have proposed different methods to tackle the bias in IR systems . These approaches include new ranking algorithms taking fairness constraints into account , post-processing method for re-ranking existing systems considering both individual and group fairness , and evaluation of ranking systems in terms of ... | {
"cite_N": [
"@cite_3"
],
"mid": [
"2544318541"
],
"abstract": [
"Ranking and scoring are ubiquitous. We consider the setting in which an institution, called a ranker, evaluates a set of individuals based on demographic, behavioral or other characteristics. The final output is a ranking that repr... |
1907.09271 | 2963401517 | Deterministic finite automata are one of the simplest and most practical models of computation studied in automata theory. Their conceptual extension is the non-deterministic finite automata which also have plenty of applications. In this article, we study these models through the lens of succinct data structures where... | The field of succinct data structures originally started with the work of Jacobson @cite_11 , and by now it is a relatively mature field in terms of breadth of problems considered. To illustrate this further, there already exists a large body of work on representing various combinatorial objects succinctly. A partial l... | {
"cite_N": [
"@cite_18",
"@cite_7",
"@cite_21",
"@cite_6",
"@cite_0",
"@cite_19",
"@cite_23",
"@cite_15",
"@cite_16",
"@cite_20",
"@cite_11"
],
"mid": [
"2147935317",
"1988115395",
"1987699222",
"2024465063",
"1993294649",
"2093788424",
"290... |
1907.09271 | 2963401517 | Deterministic finite automata are one of the simplest and most practical models of computation studied in automata theory. Their conceptual extension is the non-deterministic finite automata which also have plenty of applications. In this article, we study these models through the lens of succinct data structures where... | For DFA and NFA, other than the basic structure that is mentioned in the introduction, there exists many extensions variations in the literature, for example, two-way finite automata, B " u chi automata and many more. Researchers generally study the properties, limitations and applications of these mathematical structu... | {
"cite_N": [
"@cite_3",
"@cite_4",
"@cite_12"
],
"mid": [
"136845799",
"2404583597",
"2013433986"
],
"abstract": [
"We give asymptotic estimates and some explicit computations for both the number of distinct languages and the number of distinct finite languages over a k-letter alp... |
1907.09160 | 2964033912 | Facial MicroExpressions (MEs) are spontaneous, involuntary facial movements when a person experiences an emotion but deliberately or unconsciously attempts to conceal his or her genuine emotions. Recently, ME recognition has attracted increasing attention due to its potential applications such as clinical diagnosis, bu... | Feature representation approaches of ME recognition can be divided into two distinct categories: geometric-based and appearance-based @cite_11 methods. Specifically, geometric-based features describe the face geometry such as the shapes and locations of facial landmarks, so they need precise landmarking and alignment p... | {
"cite_N": [
"@cite_28",
"@cite_6",
"@cite_45",
"@cite_51",
"@cite_11"
],
"mid": [
"2128730107",
"2538953432",
"2139916508",
"2808103905",
"2156503193"
],
"abstract": [
"System theoretic approaches to action recognition model the dynamics of a scene with linear dyn... |
1907.09160 | 2964033912 | Facial MicroExpressions (MEs) are spontaneous, involuntary facial movements when a person experiences an emotion but deliberately or unconsciously attempts to conceal his or her genuine emotions. Recently, ME recognition has attracted increasing attention due to its potential applications such as clinical diagnosis, bu... | Since the pioneering work by Pfister @cite_37 , LBPTOP has emerged as the most popular approach for spontaneous ME analysis, and quite a few variants have been proposed. LBP Six Interception Points (LBPSIP) @cite_16 is based on three intersecting lines crossing over the center point. LBP Mean Orthogonal Planes (LBP-MOP... | {
"cite_N": [
"@cite_37",
"@cite_14",
"@cite_0",
"@cite_43",
"@cite_44",
"@cite_23",
"@cite_50",
"@cite_31",
"@cite_16",
"@cite_17"
],
"mid": [
"2059068649",
"2782644834",
"2737463186",
"234798790",
"1901075642",
"2131628350",
"2237362194",
"... |
1907.09160 | 2964033912 | Facial MicroExpressions (MEs) are spontaneous, involuntary facial movements when a person experiences an emotion but deliberately or unconsciously attempts to conceal his or her genuine emotions. Recently, ME recognition has attracted increasing attention due to its potential applications such as clinical diagnosis, bu... | @cite_35 adopts a shallow network with Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM). Other neural networks are explored in Dual Temporal Scale Convolutional Neural Network (DTSCNN) @cite_56 , 3D Flow Convolutional neural network (3DFCNN) @cite_15 and Micro-Expression Recognition algorithm using... | {
"cite_N": [
"@cite_35",
"@cite_15",
"@cite_5",
"@cite_56"
],
"mid": [
"2527254703",
"2900180747",
"2910652840",
"2760370272"
],
"abstract": [
"Recognizing spontaneous micro-expression in video sequences is a challenging problem. In this paper, we propose a new method of s... |
1907.09160 | 2964033912 | Facial MicroExpressions (MEs) are spontaneous, involuntary facial movements when a person experiences an emotion but deliberately or unconsciously attempts to conceal his or her genuine emotions. Recently, ME recognition has attracted increasing attention due to its potential applications such as clinical diagnosis, bu... | LBP was firstly proposed in @cite_41 , and a completed version was developed in @cite_9 . Later on it was introduced to face recognition in @cite_8 and its 3D extended version LBPTOP was proposed in @cite_45 with application to facial expression analysis. | {
"cite_N": [
"@cite_41",
"@cite_9",
"@cite_45",
"@cite_8"
],
"mid": [
"2039051707",
"2163352848",
"2139916508",
"2163808566"
],
"abstract": [
"This paper evaluates the performance both of some texture measures which have been successfully used in various applications and o... |
1907.09160 | 2964033912 | Facial MicroExpressions (MEs) are spontaneous, involuntary facial movements when a person experiences an emotion but deliberately or unconsciously attempts to conceal his or her genuine emotions. Recently, ME recognition has attracted increasing attention due to its potential applications such as clinical diagnosis, bu... | LBPTOP @cite_45 is the 3D extension of LBP by extracting LBP patterns separately from three orthogonal planes: the spatial plane (XY) similar to the regular LBP, the vertical spatiotemporal plane (YT) and the horizontal spatiotemporal plane (XT), as illustrated in Figure (b). | {
"cite_N": [
"@cite_45"
],
"mid": [
"2139916508"
],
"abstract": [
"Dynamic texture (DT) is an extension of texture to the temporal domain. Description and recognition of DTs have attracted growing attention. In this paper, a novel approach for recognizing DTs is proposed and its simplifications a... |
1907.09173 | 2964156559 | With the rapid development of computing technology, wearable devices such as smart phones and wristbands make it easy to get access to people's health information including activities, sleep, sports, etc. Smart healthcare achieves great success by training machine learning models on a large quantity of user data. Howev... | A comprehensive survey on federated learning is in @cite_45 . Federated machine learning was firstly proposed by Google @cite_38 @cite_38 , where they trained machine learning models based on distributed mobile phones all over the world. The key idea is to protect user data during the process. Since then, other researc... | {
"cite_N": [
"@cite_38",
"@cite_26",
"@cite_28",
"@cite_27",
"@cite_45",
"@cite_13",
"@cite_11"
],
"mid": [
"2530417694",
"",
"2767079719",
"2788629937",
"2912213068",
"",
"2777914285"
],
"abstract": [
"We introduce a new and increasingly relevant s... |
1907.09173 | 2964156559 | With the rapid development of computing technology, wearable devices such as smart phones and wristbands make it easy to get access to people's health information including activities, sleep, sports, etc. Smart healthcare achieves great success by training machine learning models on a large quantity of user data. Howev... | According to @cite_45 , federated learning can mainly be classified into three types: 1) horizontal federated learning, where organizations share partial features; 2) vertical federated learning, where organizations share partial samples; and 3) federated transfer learning, where neither samples or features have much i... | {
"cite_N": [
"@cite_45"
],
"mid": [
"2912213068"
],
"abstract": [
"Today’s artificial intelligence still faces two major challenges. One is that, in most industries, data exists in the form of isolated islands. The other is the strengthening of data privacy and security. We propose a possible sol... |
1907.09173 | 2964156559 | With the rapid development of computing technology, wearable devices such as smart phones and wristbands make it easy to get access to people's health information including activities, sleep, sports, etc. Smart healthcare achieves great success by training machine learning models on a large quantity of user data. Howev... | Transfer learning aims at transferring knowledge from existing domains to a new domain. In the setting of transfer learning, the domains are often different but related, which makes knowledge transfer possible. The key idea is to reduce the distribution divergence between different domains. To this end, there are mainl... | {
"cite_N": [
"@cite_18",
"@cite_14",
"@cite_7",
"@cite_21",
"@cite_1",
"@cite_32",
"@cite_16",
"@cite_39",
"@cite_0",
"@cite_43",
"@cite_40",
"@cite_19",
"@cite_5",
"@cite_46",
"@cite_34"
],
"mid": [
"2773817138",
"1565327149",
"2963275094",... |
1901.08933 | 2906424389 | Learning with auxiliary tasks has been shown to improve the generalisation of a primary task. However, this comes at the cost of manually-labelling additional tasks which may, or may not, be useful for the primary task. We propose a new method which automatically learns labels for an auxiliary task, such that any super... | The aim of multi-task learning (MTL) is to achieve shared representations by simultaneously training a set of related learning tasks. In this case, the learned knowledge used to share across domains is encoded into the feature representations to improve performance of each individual task, since knowledge distilled fro... | {
"cite_N": [
"@cite_4"
],
"mid": [
"2795042520"
],
"abstract": [
"In this paper, we propose a novel multi-task learning architecture, which incorporates recent advances in attention mechanisms. Our approach, the Multi-Task Attention Network (MTAN), consists of a single shared network containing a... |
1901.08933 | 2906424389 | Learning with auxiliary tasks has been shown to improve the generalisation of a primary task. However, this comes at the cost of manually-labelling additional tasks which may, or may not, be useful for the primary task. We propose a new method which automatically learns labels for an auxiliary task, such that any super... | Meta learning (or learning to learn) aims to induce the learning algorithm itself. Early works in meta learning explored automatically learning update rules for neural models . Recent approaches have focussed on learning optimisers for deep networks based on LSTMs or synthetic gradients . Meta learning has also been st... | {
"cite_N": [
"@cite_12",
"@cite_3"
],
"mid": [
"2601450892",
"2963341924"
],
"abstract": [
"A recent approach to few-shot classification called matching networks has demonstrated the benefits of coupling metric learning with a training procedure that mimics test. This approach relies on a... |
1901.09005 | 2951884559 | Unsupervised visual representation learning remains a largely unsolved problem in computer vision research. Among a big body of recently proposed approaches for unsupervised learning of visual representations, a class of self-supervised techniques achieves superior performance on many challenging benchmarks. A large nu... | In this paper we focus on self-supervised techniques that learn from image databases. These techniques have demonstrated impressive results for learning high-level image representations. Inspired by unsupervised methods from the natural language processing domain which rely on predicting words from their context @cite_... | {
"cite_N": [
"@cite_26",
"@cite_41",
"@cite_32",
"@cite_15",
"@cite_16"
],
"mid": [
"343636949",
"1614298861",
"2963465221",
"2963103975",
"2321533354"
],
"abstract": [
"This work explores the use of spatial context as a source of free and plentiful supervisory sig... |
1901.09005 | 2951884559 | Unsupervised visual representation learning remains a largely unsolved problem in computer vision research. Among a big body of recently proposed approaches for unsupervised learning of visual representations, a class of self-supervised techniques achieves superior performance on many challenging benchmarks. A large nu... | In contrast to patch-based methods, some methods generate cleverly designed image-level classification tasks. For instance, in @cite_13 Gidaris al propose to randomly rotate an image by one of four possible angles and let the model predict that rotation. Another way to create class labels is to use clustering of the im... | {
"cite_N": [
"@cite_37",
"@cite_22",
"@cite_39",
"@cite_44",
"@cite_45",
"@cite_31",
"@cite_13",
"@cite_12"
],
"mid": [
"2949532563",
"2326925005",
"2883725317",
"2842511635",
"2575671312",
"2963420272",
"2962742544",
"2750549109"
],
"abstract":... |
1901.09005 | 2951884559 | Unsupervised visual representation learning remains a largely unsolved problem in computer vision research. Among a big body of recently proposed approaches for unsupervised learning of visual representations, a class of self-supervised techniques achieves superior performance on many challenging benchmarks. A large nu... | Finally, many works have tried to combine multiple pretext tasks in one way or another. For instance, Kim al extend the jigsaw puzzle'' task by combining it with colorization and inpainting in @cite_17 . Combining the jigsaw puzzle task with clustering-based pseudo labels as in @cite_39 leads to the method called Jigsa... | {
"cite_N": [
"@cite_5",
"@cite_32",
"@cite_39",
"@cite_17"
],
"mid": [
"",
"2963465221",
"2883725317",
"2963826423"
],
"abstract": [
"",
"In self-supervised learning, one trains a model to solve a so-called pretext task on a dataset without the need for human annotatio... |
1901.09005 | 2951884559 | Unsupervised visual representation learning remains a largely unsolved problem in computer vision research. Among a big body of recently proposed approaches for unsupervised learning of visual representations, a class of self-supervised techniques achieves superior performance on many challenging benchmarks. A large nu... | The latter work is similar to ours since it contains a comparison of different self-supervision methods using a unified neural network architecture, but with the goal of combining all these tasks into a single self-supervision task. The authors use a modified ResNet101 architecture @cite_9 without further investigation... | {
"cite_N": [
"@cite_9"
],
"mid": [
"2194775991"
],
"abstract": [
"Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learnin... |
1901.09118 | 2915019010 | During active learning, an effective stopping method allows users to limit the number of annotations, which is cost effective. In this paper, a new stopping method called Predicted Change of F Measure will be introduced that attempts to provide the users an estimate of how much performance of the model is changing at e... | In 2008, Laws and Schutze proposed two stopping methods called performance convergence and uncertainty convergence @cite_37 . Performance convergence stops when the estimation of accuracy of the unlabeled pool converges so that the gradient of performance estimates is below a certain threshold. Uncertainty convergence ... | {
"cite_N": [
"@cite_37"
],
"mid": [
"1971731086"
],
"abstract": [
"Active learning is a proven method for reducing the cost of creating the training sets that are necessary for statistical NLP. However, there has been little work on stopping criteria for active learning. An operational stopping c... |
1901.09118 | 2915019010 | During active learning, an effective stopping method allows users to limit the number of annotations, which is cost effective. In this paper, a new stopping method called Predicted Change of F Measure will be introduced that attempts to provide the users an estimate of how much performance of the model is changing at e... | In 2010, Ghayoomi proposed a stopping criterion called the extended variance model @cite_16 . The extended variance model stops once the variance of the confidence of the unlabeled pool decreases by a minimum threshold over a certain number of iterations. However, in our experiments with text classification (see ), thi... | {
"cite_N": [
"@cite_16"
],
"mid": [
"1830218610"
],
"abstract": [
"Active learning is a promising method to reduce human's effort for data annotation in different NLP applications. Since it is an iterative task, it should be stopped at some point which is optimum or near-optimum. In this paper we... |
1901.09054 | 2913880062 | Two things seem to be indisputable in the contemporary deep learning discourse: 1. The categorical cross-entropy loss after softmax activation is the method of choice for classification. 2. Training a CNN classifier from scratch on small datasets does not work well. In contrast to this, we show that the cosine loss fun... | The problem of learning from limited data has been approached from various directions. First and foremost, there is a huge body of work in the field of . In this area, it is often assumed to be given a set of classes with sufficient training data that is used to improve the performance on another set of classes with ve... | {
"cite_N": [
"@cite_4",
"@cite_22",
"@cite_16",
"@cite_13",
"@cite_11"
],
"mid": [
"2768228940",
"2950537964",
"2432717477",
"",
"2786817236"
],
"abstract": [
"We present a conceptually simple, flexible, and general framework for few-shot learning, where a classifi... |
1901.09054 | 2913880062 | Two things seem to be indisputable in the contemporary deep learning discourse: 1. The categorical cross-entropy loss after softmax activation is the method of choice for classification. 2. Training a CNN classifier from scratch on small datasets does not work well. In contrast to this, we show that the cosine loss fun... | In contrast to all approaches mentioned above, our work focuses on learning from limited amounts of data without any external data or prior knowledge. This problem has recently also been tackled by incorporating a GAN for data augmentation into the learning process @cite_5 . As opposed to this, we approach the problem ... | {
"cite_N": [
"@cite_5"
],
"mid": [
"2952305675"
],
"abstract": [
"Deep learning has revolutionized the performance of classification, but meanwhile demands sufficient labeled data for training. Given insufficient data, while many techniques have been developed to help combat overfitting, the chal... |
1901.08942 | 2949384672 | We explore the use of a knowledge graphs, that capture general or commonsense knowledge, to augment the information extracted from images by the state-of-the-art methods for image captioning. The results of our experiments, on several benchmark data sets such as MS COCO, as measured by CIDEr-D, a performance metric for... | However, none of the existing methods take advantage of the readily available background knowledge about the world e.g., in the form of knowledge graphs . Such background knowledge has been shown to be useful in a broad range of applications ranging from information retrieval to question answering @cite_27 , including ... | {
"cite_N": [
"@cite_27",
"@cite_15"
],
"mid": [
"2759136286",
"2560920409"
],
"abstract": [
"Knowledge graph (KG) embedding is to embed components of a KG including entities and relations into continuous vector spaces, so as to simplify the manipulation while preserving the inherent struc... |
1901.08942 | 2949384672 | We explore the use of a knowledge graphs, that capture general or commonsense knowledge, to augment the information extracted from images by the state-of-the-art methods for image captioning. The results of our experiments, on several benchmark data sets such as MS COCO, as measured by CIDEr-D, a performance metric for... | Unlike the state-of-the-art image captioning systems, CNet-NIC is specifically designed to take advantage of background knowledge to augment the information extracted from the image (image features, objects) to improve machine-produced captions or image descriptions. Unlike VQA @cite_15 , which uses a knowledge graph t... | {
"cite_N": [
"@cite_15"
],
"mid": [
"2560920409"
],
"abstract": [
"Much of the recent progress in Vision-to-Language problems has been achieved through a combination of Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs). This approach does not explicitly represent high-leve... |
1901.09165 | 2914528260 | In this paper, we generally formulate the dynamics prediction problem of various network systems (e.g., the prediction of mobility, traffic and topology) as the temporal link prediction task. Different from conventional techniques of temporal link prediction that ignore the potential non-linear characteristics and the ... | To avoid collapsing the temporal networks, authors of @cite_21 represented the dynamic network as a third-order tensor, and the temporal information was explored by conducting a tensor factorization process. In @cite_1 , a model based on the non-negative matrix factorization (NMF) framework @cite_19 was developed, wher... | {
"cite_N": [
"@cite_19",
"@cite_21",
"@cite_1",
"@cite_22"
],
"mid": [
"",
"1864134408",
"2777190022",
"2886263965"
],
"abstract": [
"",
"The data in many disciplines such as social networks, Web analysis, etc. is link-based, and the link structure can be exploited for... |
1901.09165 | 2914528260 | In this paper, we generally formulate the dynamics prediction problem of various network systems (e.g., the prediction of mobility, traffic and topology) as the temporal link prediction task. Different from conventional techniques of temporal link prediction that ignore the potential non-linear characteristics and the ... | However, the aforementioned approaches still have limited room for the improvement of prediction accuracy, because they are almost based on the traditional linear model, ignoring the potential non-linear characteristic of the dynamic network. Although several non-linear methods based on the restricted Boltzmann machine... | {
"cite_N": [
"@cite_27",
"@cite_4"
],
"mid": [
"2782823532",
"2951508985"
],
"abstract": [
"Time varying problems usually have complex underlying structures represented as dynamic networks where entities and relationships appear and disappear over time. The problem of efficiently performi... |
1901.08522 | 2969014073 | We present an augmented reality human-swarm interface that combines two modalities of interaction: environment-oriented and robot-oriented. The environment-oriented modality allows the user to modify the environment (either virtual or physical) to indicate a goal to attain for the robot swarm. The robot-oriented modali... | Robot-oriented interactions occur when a user must engage with individual robots, e.g., to make them into leaders other robots must follow @cite_0 , to hand-pick robots for a specific task @cite_14 @cite_10 @cite_4 @cite_6 @cite_15 , or to use a robot as tangible interface for gaming and education @cite_5 @cite_17 . Th... | {
"cite_N": [
"@cite_14",
"@cite_4",
"@cite_6",
"@cite_0",
"@cite_5",
"@cite_15",
"@cite_10",
"@cite_17"
],
"mid": [
"1854667150",
"1481103847",
"2513313098",
"1590784139",
"2533619018",
"2562583785",
"2074713499",
"2577675456"
],
"abstract": [
... |
1901.08522 | 2969014073 | We present an augmented reality human-swarm interface that combines two modalities of interaction: environment-oriented and robot-oriented. The environment-oriented modality allows the user to modify the environment (either virtual or physical) to indicate a goal to attain for the robot swarm. The robot-oriented modali... | At the opposite side of the spectrum, swarm-oriented interactions occur when a user treats the swarm as a unique entity. This modality of interaction has been demonstrated in navigation tasks, e.g., beacon-based @cite_16 , density-based @cite_7 , and waypoint-based @cite_8 . The main advantage of swarm-based interactio... | {
"cite_N": [
"@cite_16",
"@cite_7",
"@cite_8"
],
"mid": [
"2118389263",
"1556924139",
"1986523056"
],
"abstract": [
"This study shows that appropriate human interaction can benefit a swarm of robots to achieve goals more efficiently. A set of desirable features for human swarm int... |
1901.08522 | 2969014073 | We present an augmented reality human-swarm interface that combines two modalities of interaction: environment-oriented and robot-oriented. The environment-oriented modality allows the user to modify the environment (either virtual or physical) to indicate a goal to attain for the robot swarm. The robot-oriented modali... | Kolling @cite_13 performed a study that is central to the topic of this paper. They compared two modalities of controlling a swarm, namely robot-oriented and environment-oriented, in a task in which the robots had to diffuse in the environment while avoiding connectivity loss. The robots performed a simple form of fora... | {
"cite_N": [
"@cite_13"
],
"mid": [
"2109664384"
],
"abstract": [
"In this paper we present the first study of human-swarm interaction comparing two fundamental types of interaction, coined intermittent and environmental. These types are exemplified by two control methods, selection and beacon co... |
1901.08544 | 2952763926 | Space partitions of @math underlie a vast and important class of fast nearest neighbor search (NNS) algorithms. Inspired by recent theoretical work on NNS for general metric spaces [Andoni, Naor, Nikolov, Razenshteyn, Waingarten STOC 2018, FOCS 2018], we develop a new framework for building space partitions reducing th... | On the empirical side, currently the fastest indexing techniques for the NNS problem are @cite_30 . The high-level idea is to construct a graph on the dataset (it can be the @math -NN graph, but other constructions are also possible), and then for each query perform a walk, which eventually converges to the nearest nei... | {
"cite_N": [
"@cite_30",
"@cite_9",
"@cite_28"
],
"mid": [
"2595294663",
"642889137",
""
],
"abstract": [
"The goal of compressed sensing is to estimate a vector from an underdetermined system of noisy linear measurements, by making use of prior knowledge on the structure of vecto... |
1901.08560 | 2911822711 | We introduce @math , an extreme case of semi-supervised learning with ultra-sparse categorisation where some classes have no labels in the training set. That is, in the training data some classes are sparsely labelled and other classes appear only as unlabelled data. Many real-world datasets are conceivably of this typ... | For clustering, both VaDE @cite_4 and GM-VAE @cite_37 extend VAEs with some form of mixture model in their learnt, continuous latent space. VaDE has the same forward model as the first model we will propose, but it uses Bayes' rule to define its classifier variational posterior over labels, rather than having a separat... | {
"cite_N": [
"@cite_37",
"@cite_4",
"@cite_7"
],
"mid": [
"2556467266",
"2730106296",
"2608412030"
],
"abstract": [
"We study a variant of the variational autoencoder model (VAE) with a Gaussian mixture as a prior distribution, with the goal of performing unsupervised clustering t... |
1901.08649 | 2912432356 | We present a novel method for learning a set of disentangled reward functions that sum to the original environment reward and are constrained to be independently obtainable. We define independent obtainability in terms of value functions with respect to obtaining one learned reward while pursuing another learned reward... | Some methods seek robust interpretable disentangled features @cite_3 @cite_2 @cite_0 @cite_4 . For example, does so by creating an information bottleneck'' @cite_7 that pressures the latent representation to be unit Gaussian. accomplishes a similar goal by maximizing the mutual information between components of their l... | {
"cite_N": [
"@cite_4",
"@cite_7",
"@cite_1",
"@cite_3",
"@cite_0",
"@cite_2"
],
"mid": [
"2737047298",
"2964184826",
"1594201624",
"1541109270",
"2810132790",
"2951392118"
],
"abstract": [
"The large pose discrepancy between two face images is one of the k... |
1901.08787 | 2950063140 | In this study, a multiple hypothesis tracking (MHT) algorithm for multi-target multi-camera tracking (MCT) with disjoint views is proposed. Our method forms track-hypothesis trees, and each branch of them represents a multi-camera track of a target that may move within a camera as well as move across cameras. Furthermo... | Single camera tracking (SCT), which tracks multiple targets in a single scene, is also called multi-object tracking (MOT). Many approaches have been proposed to improve the MOT. Track-by-detection, which optimizes a global objective function over many frames have emerged as a powerful MOT algorithm in recent years @cit... | {
"cite_N": [
"@cite_37",
"@cite_18",
"@cite_36",
"@cite_28",
"@cite_29",
"@cite_3",
"@cite_6",
"@cite_43",
"@cite_15",
"@cite_13"
],
"mid": [
"",
"2237765446",
"",
"2127021804",
"2111644456",
"",
"2016135469",
"",
"2100548006",
"2115... |
1901.08787 | 2950063140 | In this study, a multiple hypothesis tracking (MHT) algorithm for multi-target multi-camera tracking (MCT) with disjoint views is proposed. Our method forms track-hypothesis trees, and each branch of them represents a multi-camera track of a target that may move within a camera as well as move across cameras. Furthermo... | Multiple Hypothesis Tracking(MHT) was first presented in @cite_23 and is regarded as one of the earliest successful algorithm for visual tracking. MHT maintains all track hypotheses by building track-hypothesis trees whose branch represent a possible data association result(a track hypothesis). The probability of a tra... | {
"cite_N": [
"@cite_23"
],
"mid": [
"2127923214"
],
"abstract": [
"An algorithm for tracking multiple targets in a cluttered enviroment is developed. The algorithm is capable of initiating tracks, accounting for false or missing reports, and processing sets of dependent reports. As each measureme... |
1901.08787 | 2950063140 | In this study, a multiple hypothesis tracking (MHT) algorithm for multi-target multi-camera tracking (MCT) with disjoint views is proposed. Our method forms track-hypothesis trees, and each branch of them represents a multi-camera track of a target that may move within a camera as well as move across cameras. Furthermo... | Multi-camera tracking aims to establish target correspondences among observations obtained from multiple cameras so as to achieve consistent target labelling across all cameras in the camera network @cite_35 . Earlier research works in MCT only try to address tracking targets across cameras, assuming solved SCT. Howeve... | {
"cite_N": [
"@cite_35",
"@cite_1",
"@cite_20",
"@cite_39"
],
"mid": [
"2640181096",
"2573106815",
"2398802088",
"2090322660"
],
"abstract": [
"In this paper, a unified three-layer hierarchical approach for solving tracking problems in multiple non-overlapping cameras is p... |
1901.08573 | 2913266441 | We identify a trade-off between robustness and accuracy that serves as a guiding principle in the design of defenses against adversarial examples. Although this problem has been widely studied empirically, much remains unknown concerning the theory underlying this trade-off. In this work, we decompose the prediction er... | Compared with attack methods, adversarial defense methods are relatively fewer. Robust optimization based defenses are inspired by the above-mentioned attacks. Intuitively, the methods train a network by fitting its parameters to the adversarial examples: Following this framework, @cite_20 @cite_43 considered one-step ... | {
"cite_N": [
"@cite_43",
"@cite_55",
"@cite_20"
],
"mid": [
"2269778407",
"2964253222",
"2230740169"
],
"abstract": [
"Abstract We show that adversarial training of supervised learning models is in fact a robust optimization procedure. To do this, we establish a general framework ... |
1901.08573 | 2913266441 | We identify a trade-off between robustness and accuracy that serves as a guiding principle in the design of defenses against adversarial examples. Although this problem has been widely studied empirically, much remains unknown concerning the theory underlying this trade-off. In this work, we decompose the prediction er... | We mention another related line of research in adversarial defenses---relaxation based defenses. Given that the inner maximization in problem might be hard to solve due to the non-convexity nature of deep neural networks, @cite_30 and @cite_13 considered a convex outer approximation of the set of activations reachable ... | {
"cite_N": [
"@cite_30",
"@cite_14",
"@cite_33",
"@cite_2",
"@cite_58",
"@cite_13"
],
"mid": [
"2963496101",
"2963539647",
"2892354372",
"2962943487",
"2962935454",
"2963626025"
],
"abstract": [
"",
"Neural networks are vulnerable to adversarial example... |
1901.08573 | 2913266441 | We identify a trade-off between robustness and accuracy that serves as a guiding principle in the design of defenses against adversarial examples. Although this problem has been widely studied empirically, much remains unknown concerning the theory underlying this trade-off. In this work, we decompose the prediction er... | Despite a large amount of empirical works on adversarial defenses, many fundamental questions remain open in theory. There are a few preliminary explorations in recent years. @cite_47 derived upper bounds on the robustness to perturbations of any classification function, under the assumption that the data is generated ... | {
"cite_N": [
"@cite_31",
"@cite_42",
"@cite_57",
"@cite_59",
"@cite_47",
"@cite_16",
"@cite_11"
],
"mid": [
"2898193427",
"2803732607",
"2947294642",
"2964116600",
"2963849784",
"2892179671",
"2900946294"
],
"abstract": [
"Many machine learning mode... |
1901.08235 | 2913894455 | We propose a novel formulation for phase synchronization -- the statistical problem of jointly estimating alignment angles from noisy pairwise comparisons -- as a nonconvex optimization problem that enforces consistency among the pairwise comparisons in multiple frequency channels. Inspired by harmonic retrieval in sig... | Directly solving is NP-hard @cite_19 , but many convex and nonconvex methods have been proposed to find high quality approximate solutions. These include spectral and semi-definite programming (SDP) relaxations @cite_20 @cite_0 @cite_1 @cite_21 @cite_27 . An alternative approach using (GPM) is also studied @cite_24 @ci... | {
"cite_N": [
"@cite_29",
"@cite_21",
"@cite_1",
"@cite_0",
"@cite_19",
"@cite_27",
"@cite_24",
"@cite_10",
"@cite_20"
],
"mid": [
"2598300585",
"2100685794",
"2963381078",
"",
"2007104311",
"2161367205",
"2259617270",
"2962970675",
"21437039... |
1901.08235 | 2913894455 | We propose a novel formulation for phase synchronization -- the statistical problem of jointly estimating alignment angles from noisy pairwise comparisons -- as a nonconvex optimization problem that enforces consistency among the pairwise comparisons in multiple frequency channels. Inspired by harmonic retrieval in sig... | @cite_6 proposed the (NUG) SDP optimization framework for synchronization over compact Lie groups. The SDP is based on quadratically lifting the irreducible representations of the group elements, and imposing consistency among variables across frequency channels via a F 'e jer kernel; it is computationally expensive. @... | {
"cite_N": [
"@cite_6",
"@cite_11"
],
"mid": [
"2258798086",
"2536490006"
],
"abstract": [
"Let G be a compact group and let fij 2 L 2 (G). We dene the Non-Unique Games (NUG) problem as nding",
"Various alignment problems arising in cryo-electron microscopy, community detection, time ... |
1901.08422 | 2914319805 | This work addresses the challenges related to attacks on collaborative tagging systems, which often comes in a form of malicious annotations or profile injection attacks. In particular, we study various countermeasures against two types of such attacks for social tagging systems, the Overload attack and the Piggyback a... | The issue of security in tag-based RS in not new, but it has so far been mainly approached by solutions associated with spam detection. As per @cite_8 , anti-spam approaches in social networks are divided into 3 main categories: a) , such as CAPTCHAs , b) approaches which demote spam in search queries, and c) solutions... | {
"cite_N": [
"@cite_8"
],
"mid": [
"2127124926"
],
"abstract": [
"In recent years, social Web sites have become important components of the Web. With their success, however, has come a growing influx of spam. If left unchecked, spam threatens to undermine resource sharing, interactivity, and open... |
1901.08455 | 2913465187 | Pre-training of models in pruning algorithms plays an important role in pruning decision-making. We find that excessive pre-training is not necessary for pruning algorithms. According to this idea, we propose a pruning algorithm---Incremental pruning based on less training (IPLT). Compared with the traditional pruning ... | Many researchers try to construct sparse convolution kernels by pruning the weight of the network, so as to optimize the storage space occupied by the model. As early as around 1990, both @cite_12 and @cite_31 pruned the network parameters based on the second-order derivative, but this method has a high computational c... | {
"cite_N": [
"@cite_14",
"@cite_4",
"@cite_33",
"@cite_21",
"@cite_31",
"@cite_10",
"@cite_12",
"@cite_17"
],
"mid": [
"2963674932",
"2119144962",
"2460144244",
"2495425901",
"2114766824",
"2963981420",
"2125389748",
"2963000224"
],
"abstract": ... |
1901.08455 | 2913465187 | Pre-training of models in pruning algorithms plays an important role in pruning decision-making. We find that excessive pre-training is not necessary for pruning algorithms. According to this idea, we propose a pruning algorithm---Incremental pruning based on less training (IPLT). Compared with the traditional pruning ... | In the past two years, there has been a lot of work about filters pruning algorithms. Most papers use certain criteria to evaluate filters, and ultimately prune unimportant filters. In 2017, @cite_25 try to use @math to select unimportant filters. @cite_18 uses the scaling factor @math in batch normalization as an impo... | {
"cite_N": [
"@cite_18",
"@cite_0",
"@cite_19",
"@cite_2",
"@cite_25",
"@cite_20"
],
"mid": [
"2962851801",
"2553910756",
"2808217246",
"2951977814",
"",
"2786054724"
],
"abstract": [
"The deployment of deep convolutional neural networks (CNNs) in many real... |
1901.08455 | 2913465187 | Pre-training of models in pruning algorithms plays an important role in pruning decision-making. We find that excessive pre-training is not necessary for pruning algorithms. According to this idea, we propose a pruning algorithm---Incremental pruning based on less training (IPLT). Compared with the traditional pruning ... | In addition to the above papers, some researchers @cite_17 , @cite_9 have proposed algorithms that can be used to prune both parameters and filters. | {
"cite_N": [
"@cite_9",
"@cite_17"
],
"mid": [
"566555209",
"2963000224"
],
"abstract": [
"We revisit the idea of brain damage, i.e. the pruning of the coefficients of a neural network, and suggest how brain damage can be modified and used to speedup convolutional layers in ConvNets. The ... |
1901.08455 | 2913465187 | Pre-training of models in pruning algorithms plays an important role in pruning decision-making. We find that excessive pre-training is not necessary for pruning algorithms. According to this idea, we propose a pruning algorithm---Incremental pruning based on less training (IPLT). Compared with the traditional pruning ... | There is a work @cite_2 trying to combine training with pruning. In this paper, models are pruned in a soft mode. The biggest difference between us is that we actually prune some filters of models, and @cite_2 only add a mask to the parameters or filters, temporarily excluding the parameters or filters from the forward... | {
"cite_N": [
"@cite_2"
],
"mid": [
"2951977814"
],
"abstract": [
"This paper proposed a Soft Filter Pruning (SFP) method to accelerate the inference procedure of deep Convolutional Neural Networks (CNNs). Specifically, the proposed SFP enables the pruned filters to be updated when training the mo... |
1901.08150 | 2911251106 | Recently, graph neural networks have attracted great attention and achieved prominent performance in various research fields. Most of those algorithms have assumed pairwise relationships of objects of interest. However, in many real applications, the relationships between objects are in higher-order, beyond a pairwise ... | Graph Neural Network (GNN) is a methodology for learning deep models or embeddings on graph-structured data, which was first proposed by @cite_13 . One key aspect in GNN is to define the convolutional operator in the graph domain. @cite_56 firstly define convolution in the Fourier domain using the graph Laplacian matri... | {
"cite_N": [
"@cite_35",
"@cite_22",
"@cite_56",
"@cite_19",
"@cite_13"
],
"mid": [
"2964321699",
"2519887557",
"1662382123",
"637153065",
"2116341502"
],
"abstract": [
"In this work, we are interested in generalizing convolutional neural networks (CNNs) from low-d... |
1901.08150 | 2911251106 | Recently, graph neural networks have attracted great attention and achieved prominent performance in various research fields. Most of those algorithms have assumed pairwise relationships of objects of interest. However, in many real applications, the relationships between objects are in higher-order, beyond a pairwise ... | Meanwhile, some spatial algorithms directly perform convolution on the graph. For instance, @cite_14 learn different parameters for nodes with different degrees, then average the intermediate embeddings over the neighborhood structures. @cite_63 propose the PATCHY-SAN architecture, which selects a fixed-length sequence... | {
"cite_N": [
"@cite_14",
"@cite_39",
"@cite_63",
"@cite_2",
"@cite_15",
"@cite_10"
],
"mid": [
"2964113829",
"2963984147",
"2964145825",
"2788284887",
"2962767366",
"2952254971"
],
"abstract": [
"We introduce a convolutional neural network that operates dir... |
1901.08150 | 2911251106 | Recently, graph neural networks have attracted great attention and achieved prominent performance in various research fields. Most of those algorithms have assumed pairwise relationships of objects of interest. However, in many real applications, the relationships between objects are in higher-order, beyond a pairwise ... | Moreover, some other works concentrate on gate mechanism @cite_36 , skip connection @cite_64 , jumping connection @cite_4 , attention mechanism @cite_23 , sampling strategy @cite_18 @cite_54 , hierarchical representation @cite_16 , generative models @cite_51 @cite_42 , adversarial attack @cite_17 , . As a thorough revi... | {
"cite_N": [
"@cite_64",
"@cite_22",
"@cite_36",
"@cite_54",
"@cite_42",
"@cite_18",
"@cite_4",
"@cite_8",
"@cite_39",
"@cite_23",
"@cite_17",
"@cite_57",
"@cite_40",
"@cite_50",
"@cite_16",
"@cite_25",
"@cite_1",
"@cite_24",
"@cite_45",
... |
1901.08150 | 2911251106 | Recently, graph neural networks have attracted great attention and achieved prominent performance in various research fields. Most of those algorithms have assumed pairwise relationships of objects of interest. However, in many real applications, the relationships between objects are in higher-order, beyond a pairwise ... | As analyzed above, most existing variants of GNN assume pairwise relationships between objects, while our work operates on a high-order hypergraph @cite_33 @cite_6 where the between-object relationships are beyond pairwise. Hypergraph learning methods differ in the structure of the hypergraph, , clique expansion and st... | {
"cite_N": [
"@cite_35",
"@cite_26",
"@cite_33",
"@cite_6",
"@cite_3",
"@cite_27",
"@cite_46",
"@cite_12"
],
"mid": [
"2964321699",
"2086109639",
"",
"2962935106",
"2892880750",
"2148070710",
"2034204728",
"2170057991"
],
"abstract": [
"In t... |
1901.08128 | 2915044648 | Vision-based deep reinforcement learning (RL) typically obtains performance benefit by using high capacity and relatively large convolutional neural networks (CNN). However, a large network leads to higher inference costs (power, latency, silicon area, MAC count). Many inference optimizations have been developed for CN... | Distillation was proposed in @cite_2 as a method to transfer knowledge from a trained teacher classifier neural network into an untrained student network. There are various techniques to implement distillation, and here we review the version most relevant to RL. Initially, assume a high capacity teacher classifier netw... | {
"cite_N": [
"@cite_2"
],
"mid": [
"1821462560"
],
"abstract": [
"A very simple way to improve the performance of almost any machine learning algorithm is to train many different models on the same data and then to average their predictions. Unfortunately, making predictions using a whole ensembl... |
1901.08128 | 2915044648 | Vision-based deep reinforcement learning (RL) typically obtains performance benefit by using high capacity and relatively large convolutional neural networks (CNN). However, a large network leads to higher inference costs (power, latency, silicon area, MAC count). Many inference optimizations have been developed for CN... | Distillation has also proven to be useful for neuromorphic hardware design. For example, the benefits of better sample efficiency and higher student performance through distillation were combined in @cite_7 for efficient RL policy development. In this work, a high capacity policy trained with Double DQN, and represente... | {
"cite_N": [
"@cite_7",
"@cite_8"
],
"mid": [
"2893221988",
"2314470091"
],
"abstract": [
"Low precision networks in the reinforcement learning (RL) setting are relatively unexplored because of the limitations of binary activations for function approximation. Here, in the discrete action ... |
1901.08360 | 2913881544 | State-of-the-art neural networks are vulnerable to adversarial examples; they can easily misclassify inputs that are imperceptibly different than their training and test data. In this work, we establish that the use of cross-entropy loss function and the low-rank features of the training data have responsibility for th... | Differential training uses the differences between the features of the training points from opposite classes. This training scheme has been intentionally introduced to improve the dynamics of the gradient descent algorithm on the training cost function; and we consider it as using an alternative cost function in the se... | {
"cite_N": [
"@cite_14",
"@cite_3",
"@cite_12"
],
"mid": [
"",
"2157364932",
"2171590421"
],
"abstract": [
"",
"We present a method for training a similarity metric from data. The method can be used for recognition or verification applications where the number of categories is... |
1901.08215 | 2914757193 | A popular asynchronous protocol for decentralized optimization is randomized gossip where a pair of neighbors concurrently update via pairwise averaging. In practice, this creates deadlocks and is vulnerable to information delays. It can also be problematic if a node is unable to response or has only access to its priv... | parallel optimization using a master-slave architecture has been widely adopted to train models @cite_5 . In this architecture, each slave pulls the shared parameters from the master, computes its own gradient or stochastic gradient, and then pushes the gradient back to the master, where gradients from all slaves are a... | {
"cite_N": [
"@cite_5",
"@cite_15",
"@cite_1",
"@cite_22"
],
"mid": [
"2130062883",
"2164278908",
"2626580042",
"2949585412"
],
"abstract": [
"Online prediction methods are typically presented as serial algorithms running on a single processor. However, in the age of web-s... |
1901.08215 | 2914757193 | A popular asynchronous protocol for decentralized optimization is randomized gossip where a pair of neighbors concurrently update via pairwise averaging. In practice, this creates deadlocks and is vulnerable to information delays. It can also be problematic if a node is unable to response or has only access to its priv... | To overcome these issues, parallel optimization over a peer-to-peer network is an attractive alternative and allows each node to talk with only a subset of nodes. Under a connected graph, each node maintains a local copy of the training model and updates it by using its own gradient or stochastic gradient and the infor... | {
"cite_N": [
"@cite_8",
"@cite_10",
"@cite_9",
"@cite_21",
"@cite_6",
"@cite_13",
"@cite_20"
],
"mid": [
"2120293976",
"2892209803",
"1571416372",
"2962756315",
"2044212084",
"2963228337",
"2963843010"
],
"abstract": [
"The goal of decentralized opt... |
1901.08215 | 2914757193 | A popular asynchronous protocol for decentralized optimization is randomized gossip where a pair of neighbors concurrently update via pairwise averaging. In practice, this creates deadlocks and is vulnerable to information delays. It can also be problematic if a node is unable to response or has only access to its priv... | There are some algorithms for directed networks @cite_16 @cite_17 where fast nodes can only start to compute updates after waiting for slow nodes, which results in much idle time and thus makes it less efficient in large networks. The AllReduce based decentralized algorithms adopt a ring graph instead of a central node... | {
"cite_N": [
"@cite_16",
"@cite_17"
],
"mid": [
"2137435346",
"2794037585"
],
"abstract": [
"We consider distributed optimization by a collection of nodes, each having access to its own convex function, whose collective goal is to minimize the sum of the functions. The communications betw... |
1901.08215 | 2914757193 | A popular asynchronous protocol for decentralized optimization is randomized gossip where a pair of neighbors concurrently update via pairwise averaging. In practice, this creates deadlocks and is vulnerable to information delays. It can also be problematic if a node is unable to response or has only access to its priv... | decentralized parallel optimization solves the problem by breaking the synchronization in each iterate. proposed an asynchronous algorithm for undirected graphs. The seminal work @cite_24 and recent work @cite_4 focus on asynchronous coordinate descent algorithms. Recently, proposed an algorithm called AD-PSGD using st... | {
"cite_N": [
"@cite_24",
"@cite_4",
"@cite_13"
],
"mid": [
"2154834860",
"2913431899",
"2963228337"
],
"abstract": [
"We present a model for asynchronous distributed computation and then proceed to analyze the convergence of natural asynchronous distributed versions of a large cla... |
1901.08227 | 2913259183 | Recently, researchers proposed various low-precision gradient compression, for efficient communication in large-scale distributed optimization. Based on these work, we try to reduce the communication complexity from a new direction. We pursue an ideal bijective mapping between two spaces of gradient distribution, so th... | Researchers proposed protocols from other perspectives to reduce communication. A prevailing method is to average parameter occasionally, but not too frequent @cite_25 @cite_43 , or just one round of averaging over final parameters @cite_29 . If the problems require the servers to frequently synchronized, we can use an... | {
"cite_N": [
"@cite_30",
"@cite_29",
"@cite_44",
"@cite_43",
"@cite_2",
"@cite_18",
"@cite_4",
"@cite_21",
"@cite_39",
"@cite_17",
"@cite_26",
"@cite_7",
"@cite_28",
"@cite_6",
"@cite_19",
"@cite_27",
"@cite_34",
"@cite_25",
"@cite_0"
],
... |
1901.08280 | 2912135812 | Time series forecasting is a crucial component of many important applications, ranging from forecasting the stock markets to energy load prediction. The high-dimensionality, velocity and variety of the data collected in these applications pose significant and unique challenges that must be carefully addressed for each ... | This work is mainly related to time series analysis using the BoF model. An increasing number of recent works employ variants of the Bag-of-Features model to perform time series analysis, e.g., forecasting, retrieval, etc. @cite_34 , a BoF-based method was proposed for extracting discriminative representations by emplo... | {
"cite_N": [
"@cite_30",
"@cite_4",
"@cite_36",
"@cite_41",
"@cite_3",
"@cite_15",
"@cite_34",
"@cite_13"
],
"mid": [
"2024362543",
"2078474587",
"2319521425",
"1975257359",
"2896419919",
"2766117736",
"2029651301",
"2229602245"
],
"abstract": [... |
1901.08280 | 2912135812 | Time series forecasting is a crucial component of many important applications, ranging from forecasting the stock markets to energy load prediction. The high-dimensionality, velocity and variety of the data collected in these applications pose significant and unique challenges that must be carefully addressed for each ... | In contrast with @cite_3 , in this work a logistic Neural BoF formulation is used. This allows for training temporal BoF models without using any sophisticated initialization schemes and or carefully tuning any hyper-parameter, e.g., the initial scaling factor of the kernel function that was employed in @cite_3 . Furth... | {
"cite_N": [
"@cite_3"
],
"mid": [
"2896419919"
],
"abstract": [
"Time-series forecasting has various applications in a wide range of domains, e.g., forecasting stock markets using limit order book data. Limit order book data provide much richer information about the behavior of stocks than its p... |
1907.09014 | 2962851388 | Sudden changes in the dynamics of robotic tasks, such as contact with an object or the latching of a door, are often viewed as inconvenient discontinuities that make manipulation difficult. However, when these transitions are well-understood, they can be leveraged to reduce uncertainty or aid manipulation---for example... | Learning object kinematic dynamics models directly from raw visual data is a promising direction for learning object motion models. The Embed to Control (E2C) method proposed by @cite_3 uses a novel deep probabilistic generative model to convert raw image pixels into a low-dimensional latent space, in which stochastic ... | {
"cite_N": [
"@cite_10",
"@cite_12",
"@cite_3"
],
"mid": [
"2890290306",
"2963149945",
"2963430173"
],
"abstract": [
"",
"We introduce SE3-Nets which are deep neural networks designed to model and learn rigid body motion from raw point cloud data. Based only on sequences of de... |
1907.09014 | 2962851388 | Sudden changes in the dynamics of robotic tasks, such as contact with an object or the latching of a door, are often viewed as inconvenient discontinuities that make manipulation difficult. However, when these transitions are well-understood, they can be leveraged to reduce uncertainty or aid manipulation---for example... | Articulation motion models can also be seen as geometric constraints imposed on two or more rigid bodies. @cite_5 have proposed a method, , to learn geometric constraints encountered in a manipulation task from non-expert human demonstrations. @cite_7 @cite_4 developed an approach to learn geometric constraints governi... | {
"cite_N": [
"@cite_5",
"@cite_4",
"@cite_7"
],
"mid": [
"2738190501",
"",
"2785710154"
],
"abstract": [
"Learning from demonstrations has been shown to be a successful method for non-experts to teach manipulation tasks to robots. These methods typically build generative models fr... |
1907.08914 | 2963176143 | We consider the problem of locating a single facility on a vertex in a given graph based on agents' preferences, where the domain of the preferences is either single-peaked or single-dipped. Our main interest is the existence of deterministic social choice functions (SCFs) that are Pareto efficient and false-name-proof... | Moulin @cite_9 proposed generalized median voter schemes, which are the only deterministic, truthful, PE, and anonymous SCFs. Procaccia and Tennenholtz @cite_10 proposed a general framework of approximate mechanism design, which evaluates the worst case performance of truthful SCFs from the perspective of competitive r... | {
"cite_N": [
"@cite_13",
"@cite_14",
"@cite_7",
"@cite_28",
"@cite_9",
"@cite_1",
"@cite_6",
"@cite_5",
"@cite_16",
"@cite_10",
"@cite_20",
"@cite_11"
],
"mid": [
"2155975756",
"2015555360",
"1918562889",
"318376482",
"2083397025",
"28086567... |
1907.08914 | 2963176143 | We consider the problem of locating a single facility on a vertex in a given graph based on agents' preferences, where the domain of the preferences is either single-peaked or single-dipped. Our main interest is the existence of deterministic social choice functions (SCFs) that are Pareto efficient and false-name-proof... | Over the last decade, false-name-proofness has been scrutinized in various mechanism design problems @cite_27 @cite_18 @cite_25 @cite_8 , as a refinement of truthfulness for such open and anonymous environments, as the internet. Bu @cite_19 clarified a connection between false-name-proofness and population monotonicity... | {
"cite_N": [
"@cite_18",
"@cite_4",
"@cite_8",
"@cite_21",
"@cite_19",
"@cite_27",
"@cite_2",
"@cite_25",
"@cite_17"
],
"mid": [
"1788052381",
"2763993194",
"2128020465",
"2575015897",
"2140361507",
"2131402800",
"2251176627",
"",
"296381746... |
1907.08914 | 2963176143 | We consider the problem of locating a single facility on a vertex in a given graph based on agents' preferences, where the domain of the preferences is either single-peaked or single-dipped. Our main interest is the existence of deterministic social choice functions (SCFs) that are Pareto efficient and false-name-proof... | Locating a public bad has also been widely studied in both economics and computer science fields. Manjunath @cite_24 characterized truthful SCFs on an interval. @cite_23 studied the model for locating two public bads. Feigenbaum and Sethuraman @cite_29 considered the cases where single-peaked and single-dipped preferen... | {
"cite_N": [
"@cite_24",
"@cite_29",
"@cite_23"
],
"mid": [
"1965412835",
"",
"2478262890"
],
"abstract": [
"We study the problem of locating a single public good along a segment when agents have single-dipped preferences. We ask whether there are unanimous and strategy-proof rule... |
1907.08937 | 2963694998 | We introduce a conceptually simple and effective method to quantify the similarity between relations in knowledge bases. Specifically, our approach is based on the divergence between the conditional probability distributions over entity pairs. In this paper, these distributions are parameterized by a very simple neural... | As many early works devoted to psychology and linguistics, especially those works exploring semantic similarity @cite_3 @cite_25 , researchers have empirically found there are various different categorizations of semantic relations among words and contexts. For promoting research on these different semantic relations, ... | {
"cite_N": [
"@cite_41",
"@cite_48",
"@cite_1",
"@cite_3",
"@cite_19",
"@cite_40",
"@cite_34",
"@cite_25",
"@cite_17"
],
"mid": [
"35945896",
"620279967",
"1654905138",
"2103318667",
"2153579005",
"1614298861",
"36087583",
"1659833910",
"210... |
1907.08823 | 2963452950 | This paper augments the reward received by a reinforcement learning agent with potential functions in order to help the agent learn (possibly stochastic) optimal policies. We show that a potential-based reward shaping scheme is able to preserve optimality of stochastic policies, and demonstrate that the ability of an a... | Entropy regularization as a way to encourage exploration of policies during the early stages of learning was studied in @cite_14 and @cite_20 . This was used to lead a policy towards states with a high reward in levine2013guided and @cite_26 . | {
"cite_N": [
"@cite_14",
"@cite_20",
"@cite_26"
],
"mid": [
"1993411524",
"2964043796",
"2964161785"
],
"abstract": [
"Any non-associative reinforcement learning algorithm can be viewed as a method for performing function optimization through (possibly noise-corrupted) sampling of... |
1907.08823 | 2963452950 | This paper augments the reward received by a reinforcement learning agent with potential functions in order to help the agent learn (possibly stochastic) optimal policies. We show that a potential-based reward shaping scheme is able to preserve optimality of stochastic policies, and demonstrate that the ability of an a... | Static potential-based functions were shown to preserve the optimality of deterministic policies in @cite_15 . This property was extended to dynamic potential-based functions in @cite_8 . The authors of @cite_24 showed that when an agent learned a policy using Q-learning, applying PBRS at each training step was equival... | {
"cite_N": [
"@cite_8",
"@cite_36",
"@cite_24",
"@cite_0",
"@cite_15"
],
"mid": [
"2151382427",
"2202549229",
"2130750514",
"2808546214",
"1777239053"
],
"abstract": [
"Potential-based reward shaping can significantly improve the time needed to learn an optimal pol... |
1907.08823 | 2963452950 | This paper augments the reward received by a reinforcement learning agent with potential functions in order to help the agent learn (possibly stochastic) optimal policies. We show that a potential-based reward shaping scheme is able to preserve optimality of stochastic policies, and demonstrate that the ability of an a... | The use of PBRS in model-based RL was studied in @cite_6 , and for episodic RL in @cite_4 . PBRS was extended to planning in partially observable domains in @cite_28 . However, these papers only considered the finite-horizon case. In comparison, we consider the infinite horizon, discounted cost setting in this paper. | {
"cite_N": [
"@cite_28",
"@cite_4",
"@cite_6"
],
"mid": [
"2061902782",
"2620974420",
""
],
"abstract": [
"In this paper, we address the problem of suboptimal behavior during online partially observable Markov decision process (POMDP) planning caused by time constraints on plannin... |
1907.08906 | 2963652799 | In this paper, we consider the colorful @math -center problem, which is a generalization of the well-known @math -center problem. Here, we are given red and blue points in a metric space, and a coverage requirement for each color. The goal is to find the smallest radius @math , such that with @math balls of radius @mat... | The @math -means and @math -median are classic -hard clustering problems that are closely related to the @math -center problem. Like the @math -center problem, these problems have been extensively studied, resulting in different approaches guaranteeing constant factor approximations. More recently, the outlier versions... | {
"cite_N": [
"@cite_18",
"@cite_21",
"@cite_13"
],
"mid": [
"2048974928",
"2734881792",
"2767218854"
],
"abstract": [
"We consider the k-median clustering with outliers problem: Given a finite point set in a metric space and parameters k and m, we want to remove m points (called o... |
1907.08906 | 2963652799 | In this paper, we consider the colorful @math -center problem, which is a generalization of the well-known @math -center problem. Here, we are given red and blue points in a metric space, and a coverage requirement for each color. The goal is to find the smallest radius @math , such that with @math balls of radius @mat... | Facility location with outliers, which is referred to as Robust Facility Location , is a generalization of the uncapacitated Facility Location problem; various constant approximations are known for the latter problem. The Robust Facility Location problem was introduced in @cite_11 , who gave a @math -approximation. The... | {
"cite_N": [
"@cite_7",
"@cite_11"
],
"mid": [
"2052494364",
"2003719999"
],
"abstract": [
"In this article, we will formalize the method of dual fitting and the idea of factor-revealing LP. This combination is used to design and analyze two greedy algorithms for the metric uncapacitated ... |
1907.08906 | 2963652799 | In this paper, we consider the colorful @math -center problem, which is a generalization of the well-known @math -center problem. Here, we are given red and blue points in a metric space, and a coverage requirement for each color. The goal is to find the smallest radius @math , such that with @math balls of radius @mat... | A colorful version of vertex cover is studied in @cite_16 , and colorful versions of the Set Cover and Facility Location-type problems were considered in @cite_19 . In these problems, the cardinality of the cover (or its weight) shows up in the objective function, unlike in @math -center, where the number of centers ba... | {
"cite_N": [
"@cite_19",
"@cite_16"
],
"mid": [
"2890196309",
"2172398637"
],
"abstract": [
"Several algorithms with an approximation guarantee of @math are known for the Set Cover problem, where @math is the number of elements. We study a generalization of the Set Cover problem, called t... |
1907.08906 | 2963652799 | In this paper, we consider the colorful @math -center problem, which is a generalization of the well-known @math -center problem. Here, we are given red and blue points in a metric space, and a coverage requirement for each color. The goal is to find the smallest radius @math , such that with @math balls of radius @mat... | Finally, @math -center and @math -median have been generalized in an orthogonal direction, where there are additional constraints on the centers @cite_20 @cite_0 @cite_3 . Again, the issues studied in these generalizations tend to be quite different from the ones we confront here. | {
"cite_N": [
"@cite_0",
"@cite_3",
"@cite_20"
],
"mid": [
"2105773506",
"",
"1558598144"
],
"abstract": [
"In the classic k-center problem, we are given a metric graph, and the objective is to open k nodes as centers such that the maximum distance from any vertex to its closest ce... |
1907.08736 | 2962812134 | Most of privacy protection studies for textual data focus on removing explicit sensitive identifiers. However, personal writing style, as a strong indicator of the authorship, is often neglected. Recent studies on writing style anonymization can only output numeric vectors which are difficult for the recipients to inte... | Recently, differential privacy has received a lot of attention in the machine learning community. The differentially-private deep learning model @cite_38 and the deep private auto-encoder @cite_26 are designed to preserve the training data privacy. Their purpose is to guarantee that publishing the trained model does no... | {
"cite_N": [
"@cite_38",
"@cite_18",
"@cite_26",
"@cite_36",
"@cite_13"
],
"mid": [
"2473418344",
"2131621068",
"2520442116",
"2798768357",
"2085472312"
],
"abstract": [
"Machine learning techniques based on neural networks are achieving remarkable results in a wid... |
1907.08736 | 2962812134 | Most of privacy protection studies for textual data focus on removing explicit sensitive identifiers. However, personal writing style, as a strong indicator of the authorship, is often neglected. Recent studies on writing style anonymization can only output numeric vectors which are difficult for the recipients to inte... | Text generation is a trending topic in machine learning. It aims at generating a text sample with changed attributes. propose a model to change the degree of politeness while generating text. combine the with to generate a sentence with different sentiment and tense. A4NT @cite_4 is able to control the gender and age a... | {
"cite_N": [
"@cite_4"
],
"mid": [
"2767368215"
],
"abstract": [
"Text-based analysis methods allow to reveal privacy relevant author attributes such as gender, age and identify of the text's author. Such methods can compromise the privacy of an anonymous author even when the author tries to remo... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.