Unnamed: 0.1
int64
0
41k
Unnamed: 0
int64
0
41k
author
stringlengths
9
1.39k
id
stringlengths
11
18
summary
stringlengths
25
3.66k
title
stringlengths
4
258
year
int64
1.99k
2.02k
arxiv_url
stringlengths
32
39
info
stringlengths
523
3.18k
embeddings
stringlengths
16.9k
17.1k
2,000
2,000
['Sergey Levine']
1311.1761v1
Sophisticated multilayer neural networks have achieved state of the art results on multiple supervised tasks. However, successful applications of such multilayer networks to control have so far been limited largely to the perception portion of the control pipeline. In this paper, we explore the application of deep and ...
Exploring Deep and Recurrent Architectures for Optimal Control
2,013
http://arxiv.org/pdf/1311.1761v1
Title Exploring Deep Recurrent Architectures Optimal Control Summary Sophisticated multilayer neural network achieved state art result multiple supervised task However successful application multilayer network control far limited largely perception portion control pipeline paper explore application deep recurrent neura...
[-0.01812690682709217, -0.0016304963501170278, -0.012724400497972965, 0.041018854826688766, -0.007980655878782272, -0.014565364457666874, 0.02182168699800968, -0.014321783557534218, -0.01280233170837164, 0.0027682483196258545, -0.05299396067857742, -0.005680026952177286, 0.000297649297863245, 0.03269273042678833, 0.043...
2,001
2,001
['Antoine Cully', 'Jeff Clune', 'Danesh Tarapore', 'Jean-Baptiste Mouret']
1407.3501v4
As robots leave the controlled environments of factories to autonomously function in more complex, natural environments, they will have to respond to the inevitable fact that they will become damaged. However, while animals can quickly adapt to a wide variety of injuries, current robots cannot "think outside the box" t...
Robots that can adapt like animals
2,014
http://arxiv.org/pdf/1407.3501v4
Title Robots adapt like animal Summary robot leave controlled environment factory autonomously function complex natural environment respond inevitable fact become damaged However animal quickly adapt wide variety injury current robot cannot think outside box find compensatory behavior damaged limited prespecified selfs...
[0.01717059500515461, -0.0015626351814717054, -0.03809599205851555, -0.06259647011756897, 0.017703227698802948, 0.004668708890676498, 0.016614459455013275, 0.025765415281057358, 0.004476375877857208, -0.04502919688820839, 0.02251107431948185, 0.08268415927886963, -0.007880711928009987, 0.08072320371866226, 0.0362542122...
2,002
2,002
['Samuel Rönnqvist', 'Peter Sarlin']
1507.07870v1
News is a pertinent source of information on financial risks and stress factors, which nevertheless is challenging to harness due to the sparse and unstructured nature of natural text. We propose an approach based on distributional semantics and deep learning with neural networks to model and link text to a scarce set ...
Detect & Describe: Deep learning of bank stress in the news
2,015
http://arxiv.org/pdf/1507.07870v1
Title Detect Describe Deep learning bank stress news Summary News pertinent source information financial risk stress factor nevertheless challenging harness due sparse unstructured nature natural text propose approach based distributional semantics deep learning neural network model link text scarce set bank distress e...
[-0.002058073179796338, 0.08103138953447342, -0.019490301609039307, 0.04193035885691643, 0.0015634261071681976, 0.04260951653122902, 0.003340967930853367, 0.05740674212574959, 0.002434351248666644, -0.03450623154640198, 0.00831101555377245, 0.02494077943265438, -0.0042356932535767555, 0.08027864992618561, 0.00114562909...
2,003
2,003
['Keisuke Fujii', 'Kohei Nakajima']
1602.08159v2
Quantum computer has an amazing potential of fast information processing. However, realisation of a digital quantum computer is still a challenging problem requiring highly accurate controls and key application strategies. Here we propose a novel platform, quantum reservoir computing, to solve these issues successfully...
Harnessing disordered quantum dynamics for machine learning
2,016
http://arxiv.org/pdf/1602.08159v2
Title Harnessing disordered quantum dynamic machine learning Summary Quantum computer amazing potential fast information processing However realisation digital quantum computer still challenging problem requiring highly accurate control key application strategy propose novel platform quantum reservoir computing solve i...
[-0.02031875215470791, 0.04621300846338272, -0.07461051642894745, 0.03353401646018028, -0.026260945945978165, 0.0033485046587884426, -0.038478244096040726, 0.010527705773711205, 0.01091979630291462, 0.01661529764533043, 0.013808435760438442, -0.030230678617954254, -0.05104776471853256, 0.05325693637132645, 0.0055174240...
2,004
2,004
['Ivo Bukovsky', 'Noriyasu Homma']
1606.07149v1
Stability evaluation of a weight-update system of higher-order neural units (HONUs) with polynomial aggregation of neural inputs (also known as classes of polynomial neural networks) for adaptation of both feedforward and recurrent HONUs by a gradient descent method is introduced. An essential core of the approach is b...
An Approach to Stable Gradient Descent Adaptation of Higher-Order Neural Units
2,016
http://arxiv.org/pdf/1606.07149v1
Title Approach Stable Gradient Descent Adaptation HigherOrder Neural Units Summary Stability evaluation weightupdate system higherorder neural unit HONUs polynomial aggregation neural input also known class polynomial neural network adaptation feedforward recurrent HONUs gradient descent method introduced essential cor...
[-0.02515062503516674, 0.027309082448482513, -0.007191912271082401, 0.02789294719696045, 0.07024575024843216, -0.00807950273156166, -0.009035298600792885, -0.005039596930146217, 0.010862094350159168, -0.0057184211909770966, 0.01970582641661167, -0.00034858236904256046, -0.015107551589608192, 0.05062922462821007, 0.0163...
2,005
2,005
['Anna Levit', 'Daniel Crawford', 'Navid Ghadermarzy', 'Jaspreet S. Oberoi', 'Ehsan Zahedinejad', 'Pooya Ronagh']
1706.00074v1
Recent theoretical and experimental results suggest the possibility of using current and near-future quantum hardware in challenging sampling tasks. In this paper, we introduce free energy-based reinforcement learning (FERL) as an application of quantum hardware. We propose a method for processing a quantum annealer's ...
Free energy-based reinforcement learning using a quantum processor
2,017
http://arxiv.org/pdf/1706.00074v1
Title Free energybased reinforcement learning using quantum processor Summary Recent theoretical experimental result suggest possibility using current nearfuture quantum hardware challenging sampling task paper introduce free energybased reinforcement learning FERL application quantum hardware propose method processing...
[-0.029590751975774765, 0.02536092698574066, -0.039754316210746765, 0.029157575219869614, -0.03638704866170883, 0.012705642729997635, -0.042786747217178345, 0.0190141461789608, 0.016105892136693, 0.0266190804541111, -0.02405303157866001, 0.004558765795081854, -0.04941065236926079, 0.030446887016296387, 0.01874563843011...
2,006
2,006
['Qiyang Li', 'Xintong Du', 'Yizhou Huang', 'Quinlan Sykora', 'Angela P. Schoellig']
1709.06620v1
Inspired by biological swarms, robotic swarms are envisioned to solve real-world problems that are difficult for individual agents. Biological swarms can achieve collective intelligence based on local interactions and simple rules; however, designing effective distributed policies for large-scale robotic swarms to achi...
Learning of Coordination Policies for Robotic Swarms
2,017
http://arxiv.org/pdf/1709.06620v1
Title Learning Coordination Policies Robotic Swarms Summary Inspired biological swarm robotic swarm envisioned solve realworld problem difficult individual agent Biological swarm achieve collective intelligence based local interaction simple rule however designing effective distributed policy largescale robotic swarm a...
[0.04172025993466377, -0.02785640023648739, 0.020467156544327736, -0.08492074906826019, -0.006545148324221373, -0.027242621406912804, 0.040079500526189804, -0.051649708300828934, -0.0018864271696656942, 0.006202727556228638, -0.02832169085741043, 0.013419161550700665, 0.0056116352789103985, 0.022869165986776352, -0.007...
2,007
2,007
['Marius Stärk', 'Damian Backes', 'Christian Kehl']
1711.04518v1
In this article an automation system for human-machine-interfaces (HMI) for setpoint adjustment using supervised learning is presented. We use HMIs of multi-modal thermal conditioning systems in passenger cars as example for a complex setpoint selection system. The goal is the reduction of interaction complexity up to ...
A Supervised Learning Concept for Reducing User Interaction in Passenger Cars
2,017
http://arxiv.org/pdf/1711.04518v1
Title Supervised Learning Concept Reducing User Interaction Passenger Cars Summary article automation system humanmachineinterfaces HMI setpoint adjustment using supervised learning presented use HMIs multimodal thermal conditioning system passenger car example complex setpoint selection system goal reduction interacti...
[-0.051267970353364944, -0.01783495768904686, -0.04232129082083702, 0.03551000729203224, 0.019671272486448288, 0.023947130888700485, 0.07834162563085556, 0.03173733875155449, 0.016758592799305916, -0.034353941679000854, 0.006254457402974367, 0.06817091256380081, 0.03771190717816353, 0.028343942016363144, 0.004085161257...
2,008
2,008
['Nathalia Nascimento', 'Carlos Lucena', 'Paulo Alencar', 'Donald Cowan']
1802.01096v2
Several papers have recently contained reports on applying machine learning (ML) to the automation of software engineering (SE) tasks, such as project management, modeling and development. However, there appear to be no approaches comparing how software engineers fare against machine-learning algorithms as applied to s...
Software Engineers vs. Machine Learning Algorithms: An Empirical Study Assessing Performance and Reuse Tasks
2,018
http://arxiv.org/pdf/1802.01096v2
Title Software Engineers v Machine Learning Algorithms Empirical Study Assessing Performance Reuse Tasks Summary Several paper recently contained report applying machine learning ML automation software engineering SE task project management modeling development However appear approach comparing software engineer fare m...
[0.0384913794696331, 0.006360893603414297, -0.0469672828912735, 0.016465842723846436, -0.03592854365706444, -0.011814495548605919, 0.03724271059036255, -0.006641787476837635, -0.02455751597881317, -0.04381786659359932, 0.04981112852692604, 0.13979297876358032, -0.004972834140062332, 0.0583360530436039, 0.00965753756463...
2,009
2,009
['Philip Graff', 'Farhan Feroz', 'Michael P. Hobson', 'Anthony N. Lasenby']
1309.0790v2
We present the first public release of our generic neural network training algorithm, called SkyNet. This efficient and robust machine learning tool is able to train large and deep feed-forward neural networks, including autoencoders, for use in a wide range of supervised and unsupervised learning applications, such as...
SKYNET: an efficient and robust neural network training tool for machine learning in astronomy
2,013
http://arxiv.org/pdf/1309.0790v2
Title SKYNET efficient robust neural network training tool machine learning astronomy Summary present first public release generic neural network training algorithm called SkyNet efficient robust machine learning tool able train large deep feedforward neural network including autoencoders use wide range supervised unsu...
[0.0013033621944487095, 0.02770918980240822, 0.00297247595153749, 0.011962088756263256, 0.020562922582030296, 0.01774660497903824, 0.04850605130195618, 0.00537571357563138, 0.010160477831959724, 0.028806500136852264, -0.017204442992806435, 0.033716946840286255, 0.033303845673799515, 0.032450608909130096, -6.00027924519...
2,010
2,010
['Zeyuan Allen-Zhu', 'Elad Hazan']
1603.05643v2
We consider the fundamental problem in non-convex optimization of efficiently reaching a stationary point. In contrast to the convex case, in the long history of this basic problem, the only known theoretical results on first-order non-convex optimization remain to be full gradient descent that converges in $O(1/\varep...
Variance Reduction for Faster Non-Convex Optimization
2,016
http://arxiv.org/pdf/1603.05643v2
Title Variance Reduction Faster NonConvex Optimization Summary consider fundamental problem nonconvex optimization efficiently reaching stationary point contrast convex case long history basic problem known theoretical result firstorder nonconvex optimization remain full gradient descent converges O1varepsilon iteratio...
[-0.021687358617782593, 0.051756419241428375, 0.0033574197441339493, 0.03755512461066246, 0.030704880133271217, -0.026085689663887024, -0.0361269935965538, -0.009252406656742096, -0.055426280945539474, 0.012872538529336452, -0.06470102816820145, 0.01928531937301159, 0.020311487838625908, 0.06275063753128052, 0.03695965...
2,011
2,011
['Christian Beck', 'Weinan E', 'Arnulf Jentzen']
1709.05963v1
High-dimensional partial differential equations (PDE) appear in a number of models from the financial industry, such as in derivative pricing models, credit valuation adjustment (CVA) models, or portfolio optimization models. The PDEs in such applications are high-dimensional as the dimension corresponds to the number ...
Machine learning approximation algorithms for high-dimensional fully nonlinear partial differential equations and second-order backward stochastic differential equations
2,017
http://arxiv.org/pdf/1709.05963v1
Title Machine learning approximation algorithm highdimensional fully nonlinear partial differential equation secondorder backward stochastic differential equation Summary Highdimensional partial differential equation PDE appear number model financial industry derivative pricing model credit valuation adjustment CVA mod...
[-0.005092541687190533, 0.019588958472013474, -0.014310956001281738, -0.028722012415528297, 0.05115117132663727, 0.016217291355133057, 0.06456783413887024, -0.008842868730425835, -0.01795208640396595, 0.027860041707754135, 0.003746198723092675, -0.026106422767043114, -0.0382755808532238, 0.040379174053668976, 0.0211240...
2,012
2,012
['Milad Makkie', 'Heng Huang', 'Yu Zhao', 'Athanasios V. Vasilakos', 'Tianming Liu']
1710.08961v3
In recent years, analyzing task-based fMRI (tfMRI) data has become an essential tool for understanding brain function and networks. However, due to the sheer size of tfMRI data, its intrinsic complex structure, and lack of ground truth of underlying neural activities, modeling tfMRI data is hard and challenging. Previo...
Fast and Scalable Distributed Deep Convolutional Autoencoder for fMRI Big Data Analytics
2,017
http://arxiv.org/pdf/1710.08961v3
Title Fast Scalable Distributed Deep Convolutional Autoencoder fMRI Big Data Analytics Summary recent year analyzing taskbased fMRI tfMRI data become essential tool understanding brain function network However due sheer size tfMRI data intrinsic complex structure lack ground truth underlying neural activity modeling tf...
[-0.026247989386320114, 0.07213917374610901, -0.05202491581439972, 0.05725902318954468, 0.032201915979385376, 0.05146174132823944, 0.07805132865905762, 0.0352652482688427, -0.0056503708474338055, 0.04314083233475685, -0.056958772242069244, -0.026309439912438393, 0.02684960700571537, 0.09834939986467361, 0.0395484492182...
2,013
2,013
['Kamil Nar', 'Shankar Sastry']
1803.08203v1
While training error of most deep neural networks degrades as the depth of the network increases, residual networks appear to be an exception. We show that the main reason for this is the Lyapunov stability of the gradient descent algorithm: for an arbitrarily chosen step size, the equilibria of the gradient descent ar...
Residual Networks: Lyapunov Stability and Convex Decomposition
2,018
http://arxiv.org/pdf/1803.08203v1
Title Residual Networks Lyapunov Stability Convex Decomposition Summary training error deep neural network degrades depth network increase residual network appear exception show main reason Lyapunov stability gradient descent algorithm arbitrarily chosen step size equilibrium gradient descent likely remain stable param...
[-0.03477063030004501, 0.04301203414797783, -0.007012563291937113, 0.038955360651016235, 0.03203439339995384, -0.02981133945286274, -0.019885094836354256, -0.011481289751827717, -0.057549115270376205, -0.007173300255089998, -0.00899304635822773, 0.02435251511633396, -0.00628232304006815, 0.03490883484482765, -0.0050924...
2,014
2,014
['Ken Miura', 'Tatsuya Harada']
1503.05743v1
Deep learning can achieve outstanding results in various fields. However, it requires so significant computational power that graphics processing units (GPUs) and/or numerous computers are often required for the practical application. We have developed a new distributed calculation framework called "Sashimi" that allow...
Implementation of a Practical Distributed Calculation System with Browsers and JavaScript, and Application to Distributed Deep Learning
2,015
http://arxiv.org/pdf/1503.05743v1
Title Implementation Practical Distributed Calculation System Browsers JavaScript Application Distributed Deep Learning Summary Deep learning achieve outstanding result various field However requires significant computational power graphic processing unit GPUs andor numerous computer often required practical applicatio...
[-0.018067961558699608, 0.054072119295597076, -0.02249765396118164, 0.0032335149589926004, -0.03342723101377487, -0.011014917865395546, 0.11566087603569031, -0.011257179081439972, -0.03878922760486603, -0.004986946005374193, -0.023649191483855247, -0.0246924739331007, 0.05027389153838158, 0.08078402280807495, -0.010249...
2,015
2,015
['Niek Tax', 'Ilya Verenich', 'Marcello La Rosa', 'Marlon Dumas']
1612.02130v2
Predictive business process monitoring methods exploit logs of completed cases of a process in order to make predictions about running cases thereof. Existing methods in this space are tailor-made for specific prediction tasks. Moreover, their relative accuracy is highly sensitive to the dataset at hand, thus requiring...
Predictive Business Process Monitoring with LSTM Neural Networks
2,016
http://arxiv.org/pdf/1612.02130v2
Title Predictive Business Process Monitoring LSTM Neural Networks Summary Predictive business process monitoring method exploit log completed case process order make prediction running case thereof Existing method space tailormade specific prediction task Moreover relative accuracy highly sensitive dataset hand thus re...
[-0.006255712825804949, 0.006085309199988842, -0.0003740624524652958, 0.0015210873680189252, -0.027735453099012375, -0.033664148300886154, -0.02745058946311474, 0.015885265544056892, -0.016868505626916885, -0.003363007679581642, 0.017423110082745552, -0.019847236573696136, -0.02224508859217167, 0.10149114578962326, -0....
2,016
2,016
['Yasser Roudi', 'Graham Taylor']
1506.00354v2
Learning and inferring features that generate sensory input is a task continuously performed by cortex. In recent years, novel algorithms and learning rules have been proposed that allow neural network models to learn such features from natural images, written text, audio signals, etc. These networks usually involve de...
Learning with hidden variables
2,015
http://arxiv.org/pdf/1506.00354v2
Title Learning hidden variable Summary Learning inferring feature generate sensory input task continuously performed cortex recent year novel algorithm learning rule proposed allow neural network model learn feature natural image written text audio signal etc network usually involve deep architecture many layer hidden ...
[-0.0007743870373815298, -0.03279505670070648, -0.03820051625370979, 0.02613869123160839, -0.0005967774195596576, -0.00454524252563715, 0.03191288188099861, 0.02065906673669815, 0.0454266220331192, 0.007685813121497631, -0.02416374161839485, 0.07315533608198166, 0.027180617675185204, 0.04847079515457153, 0.031994044780...
2,017
2,017
['Philipp Moritz', 'Robert Nishihara', 'Ion Stoica', 'Michael I. Jordan']
1511.06051v4
Training deep networks is a time-consuming process, with networks for object recognition often requiring multiple days to train. For this reason, leveraging the resources of a cluster to speed up training is an important area of work. However, widely-popular batch-processing computational frameworks like MapReduce and ...
SparkNet: Training Deep Networks in Spark
2,015
http://arxiv.org/pdf/1511.06051v4
Title SparkNet Training Deep Networks Spark Summary Training deep network timeconsuming process network object recognition often requiring multiple day train reason leveraging resource cluster speed training important area work However widelypopular batchprocessing computational framework like MapReduce Spark designed ...
[-0.03600901737809181, -0.011357429437339306, -0.013876308687031269, 0.10761848092079163, -0.0030507389456033707, 0.010952086187899113, 0.07274629920721054, 0.006284222938120365, -0.018066421151161194, 0.01399421039968729, -0.004684088286012411, 0.04369967430830002, -0.013751113787293434, 0.06591138988733292, -0.031525...
2,018
2,018
['Miguel Á. Carreira-Perpiñán', 'Mehdi Alizadeh']
1605.09114v1
Many powerful machine learning models are based on the composition of multiple processing layers, such as deep nets, which gives rise to nonconvex objective functions. A general, recent approach to optimise such "nested" functions is the method of auxiliary coordinates (MAC). MAC introduces an auxiliary coordinate for ...
ParMAC: distributed optimisation of nested functions, with application to learning binary autoencoders
2,016
http://arxiv.org/pdf/1605.09114v1
Title ParMAC distributed optimisation nested function application learning binary autoencoders Summary Many powerful machine learning model based composition multiple processing layer deep net give rise nonconvex objective function general recent approach optimise nested function method auxiliary coordinate MAC MAC int...
[-0.011301681399345398, 0.08279135823249817, -0.029512060806155205, 0.05706649273633957, 0.00954418070614338, 0.022494150325655937, 0.09995493292808533, -0.005045757628977299, -0.03464629128575325, -0.015753133222460747, -0.03685590997338295, -0.004006894305348396, 0.013802451081573963, 0.0744948461651802, 0.0261586233...
2,019
2,019
['Raja Giryes', 'Yonina C. Eldar', 'Alex M. Bronstein', 'Guillermo Sapiro']
1605.09232v3
Solving inverse problems with iterative algorithms is popular, especially for large data. Due to time constraints, the number of possible iterations is usually limited, potentially affecting the achievable accuracy. Given an error one is willing to tolerate, an important question is whether it is possible to modify the...
Tradeoffs between Convergence Speed and Reconstruction Accuracy in Inverse Problems
2,016
http://arxiv.org/pdf/1605.09232v3
Title Tradeoffs Convergence Speed Reconstruction Accuracy Inverse Problems Summary Solving inverse problem iterative algorithm popular especially large data Due time constraint number possible iteration usually limited potentially affecting achievable accuracy Given error one willing tolerate important question whether...
[-0.014694998040795326, 0.05637459456920624, -0.005076027940958738, -0.01117612887173891, -0.02791164629161358, -0.030020011588931084, -0.01772126369178295, 0.07129532843828201, -0.03703445568680763, 0.07119801640510559, 0.0352848656475544, -0.047085702419281006, 0.006755237001925707, 0.004361252300441265, 0.0133960340...
2,020
2,020
['Dmitry Krotov', 'John J Hopfield']
1606.01164v2
A model of associative memory is studied, which stores and reliably retrieves many more patterns than the number of neurons in the network. We propose a simple duality between this dense associative memory and neural networks commonly used in deep learning. On the associative memory side of this duality, a family of mo...
Dense Associative Memory for Pattern Recognition
2,016
http://arxiv.org/pdf/1606.01164v2
Title Dense Associative Memory Pattern Recognition Summary model associative memory studied store reliably retrieves many pattern number neuron network propose simple duality dense associative memory neural network commonly used deep learning associative memory side duality family model smoothly interpolates two limiti...
[-0.017949843779206276, 0.013669214211404324, -0.042994752526283264, 0.1022338941693306, 0.011293089017271996, 0.005240985192358494, 0.04680291563272476, -0.00044335110578686, 0.0091016236692667, 0.00502577843144536, -0.010723235085606575, -0.045773450285196304, -0.0005698268068954349, 0.1003163754940033, 0.01505169831...
2,021
2,021
['Gilles Louppe', 'Michael Kagan', 'Kyle Cranmer']
1611.01046v3
Several techniques for domain adaptation have been proposed to account for differences in the distribution of the data used for training and testing. The majority of this work focuses on a binary domain label. Similar problems occur in a scientific context where there may be a continuous family of plausible data genera...
Learning to Pivot with Adversarial Networks
2,016
http://arxiv.org/pdf/1611.01046v3
Title Learning Pivot Adversarial Networks Summary Several technique domain adaptation proposed account difference distribution data used training testing majority work focus binary domain label Similar problem occur scientific context may continuous family plausible data generation process associated presence systemati...
[-0.00578206917271018, 0.030609216541051865, -0.009134912863373756, 0.03254467993974686, 0.005282165482640266, -0.037463292479515076, 0.05304005742073059, -0.008202996104955673, -0.02523788809776306, -0.008198181167244911, -0.024386510252952576, 0.010031395591795444, -0.005763475317507982, 0.08096377551555634, 0.010937...
2,022
2,022
['Benjamin Paul Chamberlain', 'Angelo Cardoso', 'C. H. Bryan Liu', 'Roberto Pagliari', 'Marc Peter Deisenroth']
1703.02596v3
We describe the Customer LifeTime Value (CLTV) prediction system deployed at ASOS.com, a global online fashion retailer. CLTV prediction is an important problem in e-commerce where an accurate estimate of future value allows retailers to effectively allocate marketing spend, identify and nurture high value customers an...
Customer Lifetime Value Prediction Using Embeddings
2,017
http://arxiv.org/pdf/1703.02596v3
Title Customer Lifetime Value Prediction Using Embeddings Summary describe Customer LifeTime Value CLTV prediction system deployed ASOScom global online fashion retailer CLTV prediction important problem ecommerce accurate estimate future value allows retailer effectively allocate marketing spend identify nurture high ...
[0.011920434422791004, 0.019795875996351242, -0.04552529752254486, 0.004753754939883947, 0.059290435165166855, 0.02248445712029934, 0.032154809683561325, -0.0207077544182539, 0.031140625476837158, -0.01666601374745369, 0.020840253680944443, 0.005243254359811544, -0.009288365021348, 0.10536667704582214, -0.0026288863737...
2,023
2,023
['Peva Blanchard', 'El Mahdi El Mhamdi', 'Rachid Guerraoui', 'Julien Stainer']
1703.02757v1
The growth of data, the need for scalability and the complexity of models used in modern machine learning calls for distributed implementations. Yet, as of today, distributed machine learning frameworks have largely ignored the possibility of arbitrary (i.e., Byzantine) failures. In this paper, we study the robustness ...
Byzantine-Tolerant Machine Learning
2,017
http://arxiv.org/pdf/1703.02757v1
Title ByzantineTolerant Machine Learning Summary growth data need scalability complexity model used modern machine learning call distributed implementation Yet today distributed machine learning framework largely ignored possibility arbitrary ie Byzantine failure paper study robustness Byzantine failure fundamental lev...
[0.014003139920532703, 0.0076744030229747295, -0.0024432498030364513, 0.0038727535866200924, -0.03820353373885155, -0.023203250020742416, 0.051640622317790985, 0.0031589900609105825, -0.006378535646945238, 0.03524745628237724, 0.02720540203154087, 0.007371662184596062, -0.00945590902119875, 0.03931954503059387, 0.00459...
2,024
2,024
['Chris Donahue', 'Zachary C. Lipton', 'Julian McAuley']
1703.06891v3
Dance Dance Revolution (DDR) is a popular rhythm-based video game. Players perform steps on a dance platform in synchronization with music as directed by on-screen step charts. While many step charts are available in standardized packs, players may grow tired of existing charts, or wish to dance to a song for which no ...
Dance Dance Convolution
2,017
http://arxiv.org/pdf/1703.06891v3
Title Dance Dance Convolution Summary Dance Dance Revolution DDR popular rhythmbased video game Players perform step dance platform synchronization music directed onscreen step chart many step chart available standardized pack player may grow tired existing chart wish dance song chart exists introduce task learning cho...
[-0.02125982753932476, -0.014263171702623367, -0.0072663212195038795, 0.021834857761859894, 0.026882274076342583, -0.019840974360704422, 0.024172013625502586, 0.006754991598427296, -0.0362471267580986, 0.029287822544574738, -0.026448026299476624, -0.009825498796999454, -0.008124668151140213, 0.06087211146950722, 0.0309...
2,025
2,025
['Ran Rubin', 'L. F. Abbott', 'Haim Sompolinsky']
1705.01502v1
Neurons and networks in the cerebral cortex must operate reliably despite multiple sources of noise. To evaluate the impact of both input and output noise, we determine the robustness of single-neuron stimulus selective responses, as well as the robustness of attractor states of networks of neurons performing memory ta...
Balanced Excitation and Inhibition are Required for High-Capacity, Noise-Robust Neuronal Selectivity
2,017
http://arxiv.org/pdf/1705.01502v1
Title Balanced Excitation Inhibition Required HighCapacity NoiseRobust Neuronal Selectivity Summary Neurons network cerebral cortex must operate reliably despite multiple source noise evaluate impact input output noise determine robustness singleneuron stimulus selective response well robustness attractor state network...
[-0.007646242622286081, -0.02206738106906414, -0.027318857610225677, 0.009025407023727894, 0.010219786316156387, -0.0004709403438027948, -0.013681769371032715, -0.022798771038651466, -0.03567575290799141, -0.0075999898836016655, -0.02909901738166809, -0.011098186485469341, 0.015837835147976875, 0.05734848603606224, 0.0...
2,026
2,026
['Weinan E', 'Jiequn Han', 'Arnulf Jentzen']
1706.04702v1
We propose a new algorithm for solving parabolic partial differential equations (PDEs) and backward stochastic differential equations (BSDEs) in high dimension, by making an analogy between the BSDE and reinforcement learning with the gradient of the solution playing the role of the policy function, and the loss functi...
Deep learning-based numerical methods for high-dimensional parabolic partial differential equations and backward stochastic differential equations
2,017
http://arxiv.org/pdf/1706.04702v1
Title Deep learningbased numerical method highdimensional parabolic partial differential equation backward stochastic differential equation Summary propose new algorithm solving parabolic partial differential equation PDEs backward stochastic differential equation BSDEs high dimension making analogy BSDE reinforcement ...
[-0.024330155923962593, 0.0095877880230546, -0.01567768305540085, -0.019662097096443176, 0.0329156294465065, -0.0012965750647708774, 0.05849071219563484, -0.009511989541351795, -0.025799810886383057, 0.014982212334871292, -0.0008000397938303649, -0.0082828588783741, -0.04176219552755356, 0.02431187592446804, 0.02580942...
2,027
2,027
['Zeyuan Allen-Zhu']
1708.08694v3
We design a stochastic algorithm to train any smooth neural network to $\varepsilon$-approximate local minima, using $O(\varepsilon^{-3.25})$ backpropagations. The best result was essentially $O(\varepsilon^{-4})$ by SGD. More broadly, it finds $\varepsilon$-approximate local minima of any smooth nonconvex function i...
Natasha 2: Faster Non-Convex Optimization Than SGD
2,017
http://arxiv.org/pdf/1708.08694v3
Title Natasha 2 Faster NonConvex Optimization SGD Summary design stochastic algorithm train smooth neural network varepsilonapproximate local minimum using Ovarepsilon325 backpropagations best result essentially Ovarepsilon4 SGD broadly find varepsilonapproximate local minimum smooth nonconvex function rate Ovarepsilon...
[-0.008425440639257431, 0.037313465029001236, 0.0021528340876102448, 0.06180174648761749, 0.017349978908896446, -0.028018075972795486, -0.019652241840958595, -0.0030623753555119038, -0.05035003274679184, 0.009007447399199009, 0.016909301280975342, -0.04319723695516586, 0.006903092376887798, 0.05348312482237816, 0.02222...
2,028
2,028
['Zeyuan Allen-Zhu', 'Yuanzhi Li']
1711.06673v2
We propose a reduction for non-convex optimization that can (1) turn an stationary-point finding algorithm into an local-minimum finding one, and (2) replace the Hessian-vector product computations with only gradient computations. It works both in the stochastic and the deterministic settings, without hurting the algor...
Neon2: Finding Local Minima via First-Order Oracles
2,017
http://arxiv.org/pdf/1711.06673v2
Title Neon2 Finding Local Minima via FirstOrder Oracles Summary propose reduction nonconvex optimization 1 turn stationarypoint finding algorithm localminimum finding one 2 replace Hessianvector product computation gradient computation work stochastic deterministic setting without hurting algorithm performance applicat...
[-0.005456711631268263, 0.01931118033826351, 0.0056770481169223785, 0.013831201009452343, 0.007601416669785976, -0.048116303980350494, 0.013809758238494396, 0.02284700982272625, -0.008668143302202225, 0.017026616260409355, 0.04537370428442955, 0.01841217279434204, 0.03634457662701607, -0.012530078180134296, -0.00762208...
2,029
2,029
['Aditya Gilra', 'Wulfram Gerstner']
1712.10158v1
Learning weights in a spiking neural network with hidden neurons, using local, stable and online rules, to control non-linear body dynamics is an open problem. Here, we employ a supervised scheme, Feedback-based Online Local Learning Of Weights (FOLLOW), to train a network of heterogeneous spiking neurons with hidden l...
Non-linear motor control by local learning in spiking neural networks
2,017
http://arxiv.org/pdf/1712.10158v1
Title Nonlinear motor control local learning spiking neural network Summary Learning weight spiking neural network hidden neuron using local stable online rule control nonlinear body dynamic open problem employ supervised scheme Feedbackbased Online Local Learning Weights FOLLOW train network heterogeneous spiking neur...
[-0.012003475800156593, 0.005737202242016792, -0.008865077048540115, 0.0027784772682935, 0.039121419191360474, -0.015537511557340622, -0.007534600794315338, -0.00042721806676127017, -0.007622255012392998, -0.03883801028132439, -0.04280460998415947, 0.0296793133020401, -0.014119190163910389, 0.07034950703382492, 0.04173...
2,030
2,030
['William H. Guss', 'Ruslan Salakhutdinov']
1802.04443v1
The learnability of different neural architectures can be characterized directly by computable measures of data complexity. In this paper, we reframe the problem of architecture selection as understanding how data determines the most expressive and generalizable architectures suited to that data, beyond inductive bias....
On Characterizing the Capacity of Neural Networks using Algebraic Topology
2,018
http://arxiv.org/pdf/1802.04443v1
Title Characterizing Capacity Neural Networks using Algebraic Topology Summary learnability different neural architecture characterized directly computable measure data complexity paper reframe problem architecture selection understanding data determines expressive generalizable architecture suited data beyond inductiv...
[-0.06849805265665054, 0.03658169135451317, -0.052782222628593445, 0.00018513103714212775, -0.06804034113883972, -0.031000711023807526, 0.008091111667454243, -0.05223986506462097, -0.016717880964279175, 0.07998142391443253, -0.025118285790085793, -0.034265220165252686, 0.006439163349568844, 0.08858127892017365, 0.05770...
2,031
2,031
['Alain Lifchitz', 'Sandra Jhean-Larose', 'Guy Denhière']
0811.0146v3
This paper presents the current state of a work in progress, whose objective is to better understand the effects of factors that significantly influence the performance of Latent Semantic Analysis (LSA). A difficult task, which consists in answering (French) biology Multiple Choice Questions, is used to test the semant...
Effect of Tuned Parameters on a LSA MCQ Answering Model
2,008
http://arxiv.org/pdf/0811.0146v3
Title Effect Tuned Parameters LSA MCQ Answering Model Summary paper present current state work progress whose objective better understand effect factor significantly influence performance Latent Semantic Analysis LSA difficult task consists answering French biology Multiple Choice Questions used test semantic property ...
[0.04870137572288513, -0.014742557890713215, -0.01124025508761406, 0.03252638131380081, 0.00900041963905096, -0.00010843017662409693, 0.009053592570126057, -0.006825895048677921, -0.07263848930597305, -0.062038227915763855, 0.0005176533595658839, 0.026962295174598694, -0.03885701298713684, -0.014458428137004375, 0.0311...
2,032
2,032
['Jonathan Huang', 'Carlos Guestrin']
1006.1328v1
Representing distributions over permutations can be a daunting task due to the fact that the number of permutations of $n$ objects scales factorially in $n$. One recent way that has been used to reduce storage complexity has been to exploit probabilistic independence, but as we argue, full independence assumptions impo...
Uncovering the Riffled Independence Structure of Rankings
2,010
http://arxiv.org/pdf/1006.1328v1
Title Uncovering Riffled Independence Structure Rankings Summary Representing distribution permutation daunting task due fact number permutation n object scale factorially n One recent way used reduce storage complexity exploit probabilistic independence argue full independence assumption impose strong sparsity constra...
[0.01070319302380085, 0.03744146600365639, -0.040114667266607285, 0.04331768676638603, -0.04185255616903305, -0.020824437960982323, 0.001161194872111082, -0.0009177264873869717, -0.05335071310400963, -0.020563025027513504, 0.00015713978791609406, 0.014739014208316803, 0.018826432526111603, -0.02269006334245205, -0.0313...
2,033
2,033
['Stephane Ross', 'Geoffrey J. Gordon', 'J. Andrew Bagnell']
1011.0686v3
Sequential prediction problems such as imitation learning, where future observations depend on previous predictions (actions), violate the common i.i.d. assumptions made in statistical learning. This leads to poor performance in theory and often in practice. Some recent approaches provide stronger guarantees in this se...
A Reduction of Imitation Learning and Structured Prediction to No-Regret Online Learning
2,010
http://arxiv.org/pdf/1011.0686v3
Title Reduction Imitation Learning Structured Prediction NoRegret Online Learning Summary Sequential prediction problem imitation learning future observation depend previous prediction action violate common iid assumption made statistical learning lead poor performance theory often practice recent approach provide stro...
[0.0204716045409441, 0.020499711856245995, 0.010235220193862915, -0.04037196561694145, -0.008327016606926918, 0.0036689976695924997, 0.007895061746239662, 0.021950384601950645, 0.0015781872207298875, -0.012559520080685616, 0.005670324433594942, -0.034390415996313095, -0.005173556040972471, 0.04169555380940437, -0.00796...
2,034
2,034
['Shakir Mohamed', 'Katherine Heller', 'Zoubin Ghahramani']
1106.1157v3
The use of L1 regularisation for sparse learning has generated immense research interest, with successful application in such diverse areas as signal acquisition, image coding, genomics and collaborative filtering. While existing work highlights the many advantages of L1 methods, in this paper we find that L1 regularis...
Bayesian and L1 Approaches to Sparse Unsupervised Learning
2,011
http://arxiv.org/pdf/1106.1157v3
Title Bayesian L1 Approaches Sparse Unsupervised Learning Summary use L1 regularisation sparse learning generated immense research interest successful application diverse area signal acquisition image coding genomics collaborative filtering existing work highlight many advantage L1 method paper find L1 regularisation o...
[-0.017403138801455498, 0.07317555695772171, -0.003959259018301964, -0.012596470303833485, 0.0026626905892044306, 0.011355162598192692, 0.06071782857179642, 0.07096890360116959, -0.06533068418502808, 0.01665109395980835, 0.03860961273312569, 0.011163071729242802, 0.06076592952013016, 0.04308663681149483, -0.00667293043...
2,035
2,035
['Stephane Ross', 'J. Andrew Bagnell']
1203.1007v2
A fundamental problem in control is to learn a model of a system from observations that is useful for controller synthesis. To provide good performance guarantees, existing methods must assume that the real system is in the class of models considered during learning. We present an iterative method with strong guarantee...
Agnostic System Identification for Model-Based Reinforcement Learning
2,012
http://arxiv.org/pdf/1203.1007v2
Title Agnostic System Identification ModelBased Reinforcement Learning Summary fundamental problem control learn model system observation useful controller synthesis provide good performance guarantee existing method must assume real system class model considered learning present iterative method strong guarantee even ...
[-0.028020884841680527, 0.026918133720755577, 0.0014182617887854576, -0.02571752853691578, 0.010957300662994385, -0.03969736769795418, 0.011495900340378284, -0.0029728510417044163, -0.03642187640070915, -0.010284488089382648, -0.006900146137923002, 0.05500250309705734, -0.01847567781805992, 0.07307399809360504, 0.01990...
2,036
2,036
['Ryan Rossi', 'Brian Gallagher', 'Jennifer Neville', 'Keith Henderson']
1203.2200v1
To understand the structural dynamics of a large-scale social, biological or technological network, it may be useful to discover behavioral roles representing the main connectivity patterns present over time. In this paper, we propose a scalable non-parametric approach to automatically learn the structural dynamics of ...
Role-Dynamics: Fast Mining of Large Dynamic Networks
2,012
http://arxiv.org/pdf/1203.2200v1
Title RoleDynamics Fast Mining Large Dynamic Networks Summary understand structural dynamic largescale social biological technological network may useful discover behavioral role representing main connectivity pattern present time paper propose scalable nonparametric approach automatically learn structural dynamic netw...
[-0.017037220299243927, -0.021922385320067406, -0.0523039847612381, -0.02144777402281761, -0.019708188250660896, -0.047278933227062225, -0.025712937116622925, -0.03269904479384422, 0.009306465275585651, 0.005421167705208063, 0.008689609356224537, 0.024475054815411568, -0.007395640481263399, 0.09341133385896683, 0.01910...
2,037
2,037
['Jonathan Sorg', 'Satinder Singh', 'Richard L. Lewis']
1203.3518v1
The explore{exploit dilemma is one of the central challenges in Reinforcement Learning (RL). Bayesian RL solves the dilemma by providing the agent with information in the form of a prior distribution over environments; however, full Bayesian planning is intractable. Planning with the mean MDP is a common myopic approxi...
Variance-Based Rewards for Approximate Bayesian Reinforcement Learning
2,012
http://arxiv.org/pdf/1203.3518v1
Title VarianceBased Rewards Approximate Bayesian Reinforcement Learning Summary exploreexploit dilemma one central challenge Reinforcement Learning RL Bayesian RL solves dilemma providing agent information form prior distribution environment however full Bayesian planning intractable Planning mean MDP common myopic app...
[0.01071979571133852, 0.028033098205924034, 0.010107972659170628, -0.07604625821113586, -0.023447997868061066, -0.016904447227716446, -0.06917952001094818, -0.00045450724428519607, -0.0422193706035614, 0.046049926429986954, -0.008258876390755177, 0.06446950882673264, -0.039577651768922806, 0.03924889117479324, -0.02098...
2,038
2,038
['Jin Tian', 'Ru He', 'Lavanya Ram']
1203.3520v1
We study the problem of learning Bayesian network structures from data. We develop an algorithm for finding the k-best Bayesian network structures. We propose to compute the posterior probabilities of hypotheses of interest by Bayesian model averaging over the k-best Bayesian networks. We present empirical results on s...
Bayesian Model Averaging Using the k-best Bayesian Network Structures
2,012
http://arxiv.org/pdf/1203.3520v1
Title Bayesian Model Averaging Using kbest Bayesian Network Structures Summary study problem learning Bayesian network structure data develop algorithm finding kbest Bayesian network structure propose compute posterior probability hypothesis interest Bayesian model averaging kbest Bayesian network present empirical res...
[-0.052136149257421494, -0.000834397622384131, -0.0029849200509488583, -0.040096938610076904, -0.018577590584754944, -0.008039669133722782, 0.02240077592432499, -0.016614142805337906, 0.012867109850049019, 0.03170822188258171, 0.04131025820970535, -0.01624329574406147, 0.03699670732021332, 0.09165886044502258, 0.012026...
2,039
2,039
['Giorgos Borboudakis', 'Ioannis Tsamardinos']
1209.6561v2
A significant theoretical advantage of search-and-score methods for learning Bayesian Networks is that they can accept informative prior beliefs for each possible network, thus complementing the data. In this paper, a method is presented for assigning priors based on beliefs on the presence or absence of certain paths ...
Scoring and Searching over Bayesian Networks with Causal and Associative Priors
2,012
http://arxiv.org/pdf/1209.6561v2
Title Scoring Searching Bayesian Networks Causal Associative Priors Summary significant theoretical advantage searchandscore method learning Bayesian Networks accept informative prior belief possible network thus complementing data paper method presented assigning prior based belief presence absence certain path true n...
[-0.028305981308221817, 0.07308406382799149, -0.009107922203838825, 0.011773981153964996, -0.03658344969153404, -0.014694617129862309, 0.001623761490918696, -0.0004610821488313377, 0.013032759539783001, 0.022091764956712723, 0.07266254723072052, 0.04407047852873802, -0.011009159497916698, 0.06972576677799225, 0.0135916...
2,040
2,040
['Tom Schaul', 'Yann LeCun']
1301.3764v2
Recent work has established an empirically successful framework for adapting learning rates for stochastic gradient descent (SGD). This effectively removes all needs for tuning, while automatically reducing learning rates over time on stationary problems, and permitting learning rates to grow appropriately in non-stati...
Adaptive learning rates and parallelization for stochastic, sparse, non-smooth gradients
2,013
http://arxiv.org/pdf/1301.3764v2
Title Adaptive learning rate parallelization stochastic sparse nonsmooth gradient Summary Recent work established empirically successful framework adapting learning rate stochastic gradient descent SGD effectively remove need tuning automatically reducing learning rate time stationary problem permitting learning rate g...
[-0.00770012941211462, 0.03009985201060772, 0.016913704574108124, 0.01749691367149353, 0.02146337740123272, 0.0042340815998613834, 0.023733124136924744, -0.03104708343744278, -0.0475950725376606, -0.009931407868862152, -0.029077570885419846, -0.010648654773831367, 0.01663900353014469, -0.03186197206377983, 0.0044883890...
2,041
2,041
['Scott Davies', 'Andrew Moore']
1301.3852v1
Recently developed techniques have made it possible to quickly learn accurate probability density functions from data in low-dimensional continuous space. In particular, mixtures of Gaussians can be fitted to data very quickly using an accelerated EM algorithm that employs multiresolution kd-trees (Moore, 1999). In thi...
Mix-nets: Factored Mixtures of Gaussians in Bayesian Networks With Mixed Continuous And Discrete Variables
2,013
http://arxiv.org/pdf/1301.3852v1
Title Mixnets Factored Mixtures Gaussians Bayesian Networks Mixed Continuous Discrete Variables Summary Recently developed technique made possible quickly learn accurate probability density function data lowdimensional continuous space particular mixture Gaussians fitted data quickly using accelerated EM algorithm empl...
[-0.05988125130534172, 0.03586592897772789, -0.027995092794299126, 0.0011349852429702878, 0.0015404297737404704, -0.004157483112066984, 0.05991163104772568, -0.009784708730876446, -0.01533148530870676, 0.05427972227334976, 0.013676410540938377, -0.0034554642625153065, 0.020802130922675133, 0.06798982620239258, 0.024268...
2,042
2,042
['Nir Friedman', 'Iftach Nachman', "Dana Pe'er"]
1301.6696v1
Learning Bayesian networks is often cast as an optimization problem, where the computational task is to find a structure that maximizes a statistically motivated score. By and large, existing learning tools address this optimization problem using standard heuristic search techniques. Since the search space is extremely...
Learning Bayesian Network Structure from Massive Datasets: The "Sparse Candidate" Algorithm
2,013
http://arxiv.org/pdf/1301.6696v1
Title Learning Bayesian Network Structure Massive Datasets Sparse Candidate Algorithm Summary Learning Bayesian network often cast optimization problem computational task find structure maximizes statistically motivated score large existing learning tool address optimization problem using standard heuristic search tech...
[-0.040849220007658005, 0.04517680034041405, -0.011775216087698936, 0.001119519816711545, -0.014401460997760296, -0.02173343487083912, 0.010028047487139702, 0.03629545867443085, -0.006224121432751417, 0.003549830289557576, 0.03286667913198471, 0.01224564015865326, 0.010467269457876682, 0.05504394695162773, -0.008310147...
2,043
2,043
['Kevin Murphy']
1301.6724v1
We show how to use a variational approximation to the logistic function to perform approximate inference in Bayesian networks containing discrete nodes with continuous parents. Essentially, we convert the logistic function to a Gaussian, which facilitates exact inference, and then iteratively adjust the variational par...
A Variational Approximation for Bayesian Networks with Discrete and Continuous Latent Variables
2,013
http://arxiv.org/pdf/1301.6724v1
Title Variational Approximation Bayesian Networks Discrete Continuous Latent Variables Summary show use variational approximation logistic function perform approximate inference Bayesian network containing discrete node continuous parent Essentially convert logistic function Gaussian facilitates exact inference iterati...
[-0.02868802472949028, 0.07039254158735275, -0.02848905324935913, -0.03319445997476578, -0.02847660332918167, -0.03733184561133385, 0.03213633596897125, -0.02282426506280899, -0.06048077344894409, 0.05302250012755394, 0.04888530820608139, 0.022066308185458183, 0.011386001482605934, 0.06789471209049225, 0.02160042896866...
2,044
2,044
['Trong Nghia Hoang', 'Kian Hsiang Low']
1304.2024v3
Recent advances in Bayesian reinforcement learning (BRL) have shown that Bayes-optimality is theoretically achievable by modeling the environment's latent dynamics using Flat-Dirichlet-Multinomial (FDM) prior. In self-interested multi-agent environments, the transition dynamics are mainly controlled by the other agent'...
A General Framework for Interacting Bayes-Optimally with Self-Interested Agents using Arbitrary Parametric Model and Model Prior
2,013
http://arxiv.org/pdf/1304.2024v3
Title General Framework Interacting BayesOptimally SelfInterested Agents using Arbitrary Parametric Model Model Prior Summary Recent advance Bayesian reinforcement learning BRL shown Bayesoptimality theoretically achievable modeling environment latent dynamic using FlatDirichletMultinomial FDM prior selfinterested mult...
[-0.008783993311226368, -0.013359913602471352, 0.01246991939842701, -0.06280293315649033, -0.00895239133387804, 0.01078389398753643, 0.005427168216556311, -0.03246433660387993, -0.03030364029109478, -0.02517791837453842, -0.04702572897076607, 0.03297744691371918, -0.05547180771827698, 0.022714046761393547, 0.0010580907...
2,045
2,045
['Teppo Niinimaki', 'Pekka Parviainen']
1210.4888v1
Learning a Bayesian network structure from data is an NP-hard problem and thus exact algorithms are feasible only for small data sets. Therefore, network structures for larger networks are usually learned with various heuristics. Another approach to scaling up the structure learning is local learning. In local learning...
Local Structure Discovery in Bayesian Networks
2,012
http://arxiv.org/pdf/1210.4888v1
Title Local Structure Discovery Bayesian Networks Summary Learning Bayesian network structure data NPhard problem thus exact algorithm feasible small data set Therefore network structure larger network usually learned various heuristic Another approach scaling structure learning local learning local learning modeler on...
[-0.03021683730185032, 0.01832403801381588, -0.021297162398695946, 0.014605916105210781, -0.03379257768392563, -0.04351023957133293, 0.019156308844685555, 0.0197920985519886, 0.05640349164605141, 0.03279845044016838, 0.057172149419784546, 0.0645824670791626, 0.008599729277193546, 0.05661757290363312, 0.0187212191522121...
2,046
2,046
['Khaled S. Refaat', 'Arthur Choi', 'Adnan Darwiche']
1210.4910v1
EDML is a recently proposed algorithm for learning MAP parameters in Bayesian networks. In this paper, we present a number of new advances and insights on the EDML algorithm. First, we provide the multivalued extension of EDML, originally proposed for Bayesian networks over binary variables. Next, we identify a simplif...
New Advances and Theoretical Insights into EDML
2,012
http://arxiv.org/pdf/1210.4910v1
Title New Advances Theoretical Insights EDML Summary EDML recently proposed algorithm learning MAP parameter Bayesian network paper present number new advance insight EDML algorithm First provide multivalued extension EDML originally proposed Bayesian network binary variable Next identify simplified characterization ED...
[-0.05227961763739586, -0.01664128713309765, -0.0218343548476696, -0.04614487662911415, -0.006675169337540865, 0.0207839272916317, -0.012842217460274696, 0.012745200656354427, 0.0030856046359986067, -0.008536397479474545, 0.04923897981643677, 0.027395859360694885, 0.03968125954270363, 0.02879505604505539, 0.01776001043...
2,047
2,047
['Thomas J. Walsh', 'Sergiu Goschin']
1210.4918v1
We describe theoretical bounds and a practical algorithm for teaching a model by demonstration in a sequential decision making environment. Unlike previous efforts that have optimized learners that watch a teacher demonstrate a static policy, we focus on the teacher as a decision maker who can dynamically choose differ...
Dynamic Teaching in Sequential Decision Making Environments
2,012
http://arxiv.org/pdf/1210.4918v1
Title Dynamic Teaching Sequential Decision Making Environments Summary describe theoretical bound practical algorithm teaching model demonstration sequential decision making environment Unlike previous effort optimized learner watch teacher demonstrate static policy focus teacher decision maker dynamically choose diffe...
[0.013284357264637947, 0.005704076960682869, -0.03142569214105606, -0.06324663013219833, -0.021392105147242546, -0.03174487501382828, 0.025764698162674904, -0.018547723069787025, -0.04425729811191559, -0.037642017006874084, 0.049486808478832245, 0.025801973417401314, 0.004723363555967808, 0.06774502247571945, -0.019277...
2,048
2,048
['Chris Häusler', 'Alex Susemihl']
1210.8353v1
Much work has been done refining and characterizing the receptive fields learned by deep learning algorithms. A lot of this work has focused on the development of Gabor-like filters learned when enforcing sparsity constraints on a natural image dataset. Little work however has investigated how these filters might expan...
Temporal Autoencoding Restricted Boltzmann Machine
2,012
http://arxiv.org/pdf/1210.8353v1
Title Temporal Autoencoding Restricted Boltzmann Machine Summary Much work done refining characterizing receptive field learned deep learning algorithm lot work focused development Gaborlike filter learned enforcing sparsity constraint natural image dataset Little work however investigated filter might expand temporal ...
[0.005382750649005175, -0.0022551564034074545, -6.478667637566105e-05, 0.03449881821870804, -0.009466838091611862, -0.01197083294391632, 0.031158292666077614, 0.010948246344923973, -0.059459783136844635, 0.011517064645886421, 0.00302535155788064, -0.024759940803050995, 0.01255862694233656, 0.1229717880487442, 0.0056922...
2,049
2,049
['Greg Ver Steeg', 'Aram Galstyan']
1406.1222v2
We introduce a method to learn a hierarchy of successively more abstract representations of complex data based on optimizing an information-theoretic objective. Intuitively, the optimization searches for a set of latent factors that best explain the correlations in the data as measured by multivariate mutual informatio...
Discovering Structure in High-Dimensional Data Through Correlation Explanation
2,014
http://arxiv.org/pdf/1406.1222v2
Title Discovering Structure HighDimensional Data Correlation Explanation Summary introduce method learn hierarchy successively abstract representation complex data based optimizing informationtheoretic objective Intuitively optimization search set latent factor best explain correlation data measured multivariate mutual...
[-0.004075735807418823, 0.06990640610456467, -0.057022932916879654, 0.02308002859354019, -0.004711890127509832, 0.03467831388115883, -0.0017232026439160109, 0.01790984347462654, 0.019846005365252495, 0.032329827547073364, -0.01767141744494438, 0.01088652666658163, 0.041693441569805145, 0.0285179503262043, -0.0083861863...
2,050
2,050
['Fajwel Fogel', "Alexandre d'Aspremont", 'Milan Vojnovic']
1406.5370v4
We describe a seriation algorithm for ranking a set of items given pairwise comparisons between these items. Intuitively, the algorithm assigns similar rankings to items that compare similarly with all others. It does so by constructing a similarity matrix from pairwise comparisons, using seriation methods to reorder t...
Spectral Ranking using Seriation
2,014
http://arxiv.org/pdf/1406.5370v4
Title Spectral Ranking using Seriation Summary describe seriation algorithm ranking set item given pairwise comparison item Intuitively algorithm assigns similar ranking item compare similarly others constructing similarity matrix pairwise comparison using seriation method reorder matrix construct ranking first show sp...
[-0.019422218203544617, 0.004632118623703718, -0.017455806955695152, 0.028462979942560196, -0.008136026561260223, -0.015131715685129166, -0.0028846978675574064, -0.02642645128071308, 9.103777119889855e-05, -0.06501538306474686, -0.04397757351398468, 0.07360626757144928, 0.030245468020439148, 0.01800582744181156, -0.000...
2,051
2,051
['Shiliang Sun', 'John Shawe-Taylor', 'Liang Mao']
1406.5614v2
This paper presents eight PAC-Bayes bounds to analyze the generalization performance of multi-view classifiers. These bounds adopt data dependent Gaussian priors which emphasize classifiers with high view agreements. The center of the prior for the first two bounds is the origin, while the center of the prior for the t...
PAC-Bayes Analysis of Multi-view Learning
2,014
http://arxiv.org/pdf/1406.5614v2
Title PACBayes Analysis Multiview Learning Summary paper present eight PACBayes bound analyze generalization performance multiview classifier bound adopt data dependent Gaussian prior emphasize classifier high view agreement center prior first two bound origin center prior third fourth bound given data dependent vector...
[-0.055217888206243515, 0.030909419059753418, 0.014141872525215149, 0.026803700253367424, 0.012504120357334614, 0.019316652789711952, 0.056129101663827896, 0.00850064866244793, 0.005768669303506613, -0.018533874303102493, -0.04060405492782593, -0.04427818953990936, 0.02400130219757557, 0.02094162628054619, 0.0215490665...
2,052
2,052
['Yuan Shi', 'Yung-Kyun Noh', 'Fei Sha', 'Daniel D. Lee']
1109.3940v1
Metrics specifying distances between data points can be learned in a discriminative manner or from generative models. In this paper, we show how to unify generative and discriminative learning of metrics via a kernel learning framework. Specifically, we learn local metrics optimized from parametric generative models. T...
Learning Discriminative Metrics via Generative Models and Kernel Learning
2,011
http://arxiv.org/pdf/1109.3940v1
Title Learning Discriminative Metrics via Generative Models Kernel Learning Summary Metrics specifying distance data point learned discriminative manner generative model paper show unify generative discriminative learning metric via kernel learning framework Specifically learn local metric optimized parametric generati...
[-0.01765531674027443, 0.0027345207054167986, -0.023071138188242912, 0.011719121597707272, -0.007242303807288408, 0.0064403279684484005, 0.024011559784412384, 0.0433473065495491, 0.0208155307918787, 0.014718227088451385, 0.020846717059612274, 0.02354065701365471, -0.006835277657955885, -0.02000403217971325, 0.017873737...
2,053
2,053
['Ilya Shpitser', 'Robin J. Evans', 'Thomas S. Richardson', 'James M. Robins']
1309.6863v1
Hidden variables are ubiquitous in practical data analysis, and therefore modeling marginal densities and doing inference with the resulting models is an important problem in statistics, machine learning, and causal inference. Recently, a new type of graphical model, called the nested Markov model, was developed which ...
Sparse Nested Markov models with Log-linear Parameters
2,013
http://arxiv.org/pdf/1309.6863v1
Title Sparse Nested Markov model Loglinear Parameters Summary Hidden variable ubiquitous practical data analysis therefore modeling marginal density inference resulting model important problem statistic machine learning causal inference Recently new type graphical model called nested Markov model developed capture equa...
[-0.007434367667883635, 0.03834656625986099, -0.006951178889721632, -0.04084799438714981, -0.028980109840631485, -0.016734736040234566, 0.046654053032398224, -0.000982991186901927, -0.024022603407502174, -0.06023741140961647, 0.07375533133745193, 0.05632682517170906, 0.019972240552306175, 0.04027143865823746, 0.0024796...
2,054
2,054
['Charmgil Hong', 'Milos Hauskrecht']
1505.04097v1
Outlier detection aims to identify unusual data instances that deviate from expected patterns. The outlier detection is particularly challenging when outliers are context dependent and when they are defined by unusual combinations of multiple outcome variable values. In this paper, we develop and study a new conditiona...
MCODE: Multivariate Conditional Outlier Detection
2,015
http://arxiv.org/pdf/1505.04097v1
Title MCODE Multivariate Conditional Outlier Detection Summary Outlier detection aim identify unusual data instance deviate expected pattern outlier detection particularly challenging outlier context dependent defined unusual combination multiple outcome variable value paper develop study new conditional outlier detect...
[-0.003778250655159354, -0.012936602346599102, -0.01818542368710041, 0.0225718654692173, -0.03248757869005203, 0.026987699791789055, 0.0458931066095829, 0.029318390414118767, -0.03590898588299751, 0.016924424096941948, 0.060610100626945496, 0.05337497219443321, 0.006394877098500729, 0.07202494889497757, -0.025095148012...
2,055
2,055
['Stephen H. Bach', 'Matthias Broecheler', 'Bert Huang', 'Lise Getoor']
1505.04406v3
A fundamental challenge in developing high-impact machine learning technologies is balancing the need to model rich, structured domains with the ability to scale to big data. Many important problem areas are both richly structured and large scale, from social and biological networks, to knowledge graphs and the Web, to...
Hinge-Loss Markov Random Fields and Probabilistic Soft Logic
2,015
http://arxiv.org/pdf/1505.04406v3
Title HingeLoss Markov Random Fields Probabilistic Soft Logic Summary fundamental challenge developing highimpact machine learning technology balancing need model rich structured domain ability scale big data Many important problem area richly structured large scale social biological network knowledge graph Web image v...
[-0.002434662077575922, 0.042974330484867096, 0.006506812758743763, -0.014104757457971573, -0.022258738055825233, -0.008792676962912083, -0.0073624905198812485, 0.0037354619707912207, 0.019013341516256332, -0.07326754182577133, 0.04684040695428848, 0.02311481535434723, 0.008768096566200256, 0.07445517927408218, -0.0009...
2,056
2,056
['Stefano Teso', 'Andrea Passerini', 'Paolo Viappiani']
1604.06020v1
In this paper we propose an approach to preference elicitation that is suitable to large configuration spaces beyond the reach of existing state-of-the-art approaches. Our setwise max-margin method can be viewed as a generalization of max-margin learning to sets, and can produce a set of "diverse" items that can be use...
Constructive Preference Elicitation by Setwise Max-margin Learning
2,016
http://arxiv.org/pdf/1604.06020v1
Title Constructive Preference Elicitation Setwise Maxmargin Learning Summary paper propose approach preference elicitation suitable large configuration space beyond reach existing stateoftheart approach setwise maxmargin method viewed generalization maxmargin learning set produce set diverse item used ask informative q...
[-0.01963682845234871, 0.016260959208011627, -0.012856283225119114, -0.03277156874537468, -0.016206320375204086, -0.022943973541259766, 0.02647341974079609, -0.0019643844570964575, -0.023047098889946938, -0.024210579693317413, 0.04536476731300354, 0.02625657431781292, 0.005023554433137178, 0.07025488466024399, -0.00131...
2,057
2,057
['Nils Y. Hammerla', 'Shane Halloran', 'Thomas Ploetz']
1604.08880v1
Human activity recognition (HAR) in ubiquitous computing is beginning to adopt deep learning to substitute for well-established analysis techniques that rely on hand-crafted feature extraction and classification techniques. From these isolated applications of custom deep architectures it is, however, difficult to gain ...
Deep, Convolutional, and Recurrent Models for Human Activity Recognition using Wearables
2,016
http://arxiv.org/pdf/1604.08880v1
Title Deep Convolutional Recurrent Models Human Activity Recognition using Wearables Summary Human activity recognition HAR ubiquitous computing beginning adopt deep learning substitute wellestablished analysis technique rely handcrafted feature extraction classification technique isolated application custom deep archi...
[-0.020750150084495544, -0.018264414742588997, -0.00896103959530592, 0.06028631329536438, 0.0240479726344347, 0.011981155723333359, 0.022620057687163353, -0.010791003704071045, -0.009540437720716, -0.038857512176036835, 0.008673683740198612, 0.03051610104739666, 0.006606362760066986, 0.01852465234696865, 0.027634572237...
2,058
2,058
['Hoifung Poon', 'Pedro Domingos']
1202.3732v1
The key limiting factor in graphical model inference and learning is the complexity of the partition function. We thus ask the question: what are general conditions under which the partition function is tractable? The answer leads to a new kind of deep architecture, which we call sum-product networks (SPNs). SPNs are d...
Sum-Product Networks: A New Deep Architecture
2,012
http://arxiv.org/pdf/1202.3732v1
Title SumProduct Networks New Deep Architecture Summary key limiting factor graphical model inference learning complexity partition function thus ask question general condition partition function tractable answer lead new kind deep architecture call sumproduct network SPNs SPNs directed acyclic graph variable leaf sum ...
[-0.025210419669747353, 0.045158471912145615, -0.013366068713366985, 0.030322378501296043, -0.012486553750932217, -0.03760001063346863, 0.005644751712679863, -0.05246814340353012, -0.002126407576724887, 0.04376307502388954, -0.016252299770712852, 0.011440097354352474, 0.03995799645781517, 0.10906574130058289, 0.0525409...
2,059
2,059
['Jonathan Huang', 'Ashish Kapoor', 'Carlos E. Guestrin']
1202.3734v1
Distributions over rankings are used to model data in various settings such as preference analysis and political elections. The factorial size of the space of rankings, however, typically forces one to make structural assumptions, such as smoothness, sparsity, or probabilistic independence about these underlying distri...
Efficient Probabilistic Inference with Partial Ranking Queries
2,012
http://arxiv.org/pdf/1202.3734v1
Title Efficient Probabilistic Inference Partial Ranking Queries Summary Distributions ranking used model data various setting preference analysis political election factorial size space ranking however typically force one make structural assumption smoothness sparsity probabilistic independence underlying distribution ...
[0.03145304694771767, 0.07926204055547714, -0.011685600504279137, -0.005350549705326557, -0.0060411738231778145, -0.018904583528637886, -0.009879871271550655, -0.023667223751544952, 0.03306824341416359, -0.02620006538927555, -0.010179316624999046, -0.0037043076008558273, -0.008397730998694897, 0.03748467564582825, -0.0...
2,060
2,060
['Alex Kulesza', 'Ben Taskar']
1202.3738v1
Determinantal point processes (DPPs), which arise in random matrix theory and quantum physics, are natural models for subset selection problems where diversity is preferred. Among many remarkable properties, DPPs offer tractable algorithms for exact inference, including computing marginal probabilities and sampling; ho...
Learning Determinantal Point Processes
2,012
http://arxiv.org/pdf/1202.3738v1
Title Learning Determinantal Point Processes Summary Determinantal point process DPPs arise random matrix theory quantum physic natural model subset selection problem diversity preferred Among many remarkable property DPPs offer tractable algorithm exact inference including computing marginal probability sampling howev...
[-0.022856617346405983, 0.007346586789935827, -0.013498600572347641, 0.006184756755828857, -0.015354282222688198, 0.013441181741654873, 0.017368970438838005, 0.011634251102805138, -0.021155428141355515, -0.0047661615535616875, 0.03468133509159088, -0.0019081940408796072, 0.024385901167988777, 0.032777294516563416, 0.02...
2,061
2,061
['Kareem Amin', 'Michael Kearns', 'Umar Syed']
1202.3782v1
We introduce a rich class of graphical models for multi-armed bandit problems that permit both the state or context space and the action space to be very large, yet succinctly specify the payoffs for any context-action pair. Our main result is an algorithm for such models whose regret is bounded by the number of parame...
Graphical Models for Bandit Problems
2,012
http://arxiv.org/pdf/1202.3782v1
Title Graphical Models Bandit Problems Summary introduce rich class graphical model multiarmed bandit problem permit state context space action space large yet succinctly specify payoff contextaction pair main result algorithm model whose regret bounded number parameter whose running time depends treewidth graph substr...
[-0.011902510188519955, 0.03232288360595703, -0.022705627605319023, -0.009746667928993702, -0.03318743407726288, -0.02093575708568096, -0.013751477003097534, 0.024013187736272812, 0.014462350867688656, -0.016177771613001823, 0.023976769298315048, 0.03734128177165985, -0.07730644941329956, 0.045788541436195374, 0.021170...
2,062
2,062
['Patrik O. Hoyer', 'Aapo Hyvarinen', 'Richard Scheines', 'Peter L. Spirtes', 'Joseph Ramsey', 'Gustavo Lacerda', 'Shohei Shimizu']
1206.3260v1
An important task in data analysis is the discovery of causal relationships between observed variables. For continuous-valued data, linear acyclic causal models are commonly used to model the data-generating process, and the inference of such models is a well-studied problem. However, existing methods have significant ...
Causal discovery of linear acyclic models with arbitrary distributions
2,012
http://arxiv.org/pdf/1206.3260v1
Title Causal discovery linear acyclic model arbitrary distribution Summary important task data analysis discovery causal relationship observed variable continuousvalued data linear acyclic causal model commonly used model datagenerating process inference model wellstudied problem However existing method significant lim...
[-0.04214766249060631, 0.04806342348456383, -0.03522440418601036, -0.006200891919434071, -0.02446010150015354, -0.02124224416911602, 0.03205832839012146, 0.01020841859281063, 0.023899681866168976, -0.004309259355068207, 0.07627962529659271, 0.0442030243575573, 0.00038116256473585963, 0.029877610504627228, -0.0149927064...
2,063
2,063
['Mauricio Araya', 'Olivier Buffet', 'Vincent Thomas']
1206.4613v1
Model-based Bayesian Reinforcement Learning (BRL) allows a found formalization of the problem of acting optimally while facing an unknown environment, i.e., avoiding the exploration-exploitation dilemma. However, algorithms explicitly addressing BRL suffer from such a combinatorial explosion that a large body of work r...
Near-Optimal BRL using Optimistic Local Transitions
2,012
http://arxiv.org/pdf/1206.4613v1
Title NearOptimal BRL using Optimistic Local Transitions Summary Modelbased Bayesian Reinforcement Learning BRL allows found formalization problem acting optimally facing unknown environment ie avoiding explorationexploitation dilemma However algorithm explicitly addressing BRL suffer combinatorial explosion large body...
[-0.002042601350694895, -0.0235260222107172, 0.013260355219244957, -0.06716302782297134, -0.03190411999821663, -0.039793070405721664, -0.04904940351843834, 0.015630632638931274, -0.0361904501914978, 0.005787061993032694, 0.029591428115963936, 0.033531304448843, -0.057629164308309555, 0.06296005845069885, -0.00664622895...
2,064
2,064
['Sergey Levine', 'Vladlen Koltun']
1206.4617v1
Inverse optimal control, also known as inverse reinforcement learning, is the problem of recovering an unknown reward function in a Markov decision process from expert demonstrations of the optimal policy. We introduce a probabilistic inverse optimal control algorithm that scales gracefully with task dimensionality, an...
Continuous Inverse Optimal Control with Locally Optimal Examples
2,012
http://arxiv.org/pdf/1206.4617v1
Title Continuous Inverse Optimal Control Locally Optimal Examples Summary Inverse optimal control also known inverse reinforcement learning problem recovering unknown reward function Markov decision process expert demonstration optimal policy introduce probabilistic inverse optimal control algorithm scale gracefully ta...
[-0.004409049171954393, 0.010758919641375542, -0.000761316332500428, -0.010187276639044285, -0.03114534728229046, -0.04473568871617317, -0.05191546306014061, 0.05069023743271828, 0.001859494368545711, 0.05494406446814537, -0.009014302864670753, 0.030578630045056343, -0.0327632762491703, 0.061896760016679764, 0.05456208...
2,065
2,065
['Kiri Wagstaff']
1206.4656v1
Much of current machine learning (ML) research has lost its connection to problems of import to the larger world of science and society. From this perspective, there exist glaring limitations in the data sets we investigate, the metrics we employ for evaluation, and the degree to which results are communicated back to ...
Machine Learning that Matters
2,012
http://arxiv.org/pdf/1206.4656v1
Title Machine Learning Matters Summary Much current machine learning ML research lost connection problem import larger world science society perspective exist glaring limitation data set investigate metric employ evaluation degree result communicated back originating domain change needed conduct research increase impac...
[0.026455124840140343, 0.042471930384635925, -0.04690465331077576, -0.009174716658890247, -0.016737869009375572, 0.008635343052446842, 0.037491075694561005, 0.004778948146849871, -0.002938994439318776, -0.03280914947390556, 0.05348948761820793, 0.06884244829416275, 0.0028328460175544024, 0.05786873400211334, 0.03556949...
2,066
2,066
['Jose M. Pena']
1206.5263v1
We present a graphical criterion for reading dependencies from the minimal directed independence map G of a graphoid p when G is a polytree and p satisfies composition and weak transitivity. We prove that the criterion is sound and complete. We argue that assuming composition and weak transitivity is not too restrictiv...
Reading Dependencies from Polytree-Like Bayesian Networks
2,012
http://arxiv.org/pdf/1206.5263v1
Title Reading Dependencies PolytreeLike Bayesian Networks Summary present graphical criterion reading dependency minimal directed independence map G graphoid p G polytree p satisfies composition weak transitivity prove criterion sound complete argue assuming composition weak transitivity restrictive Authors 0 Ahmed Osm...
[-0.018989702686667442, 0.02930963784456253, -0.028671083971858025, -0.021597659215331078, -0.043100304901599884, -0.037443604320287704, -0.021865012124180794, -0.00480780191719532, 0.009940741583704948, -0.0007444592192769051, 0.057859405875205994, 0.03750655800104141, -0.010069064795970917, 0.08158840984106064, -0.04...
2,067
2,067
['Marina Meila', 'Kapil Phadnis', 'Arthur Patterson', 'Jeff A. Bilmes']
1206.5265v1
We analyze the generalized Mallows model, a popular exponential model over rankings. Estimating the central (or consensus) ranking from data is NP-hard. We obtain the following new results: (1) We show that search methods can estimate both the central ranking pi0 and the model parameters theta exactly. The search is n!...
Consensus ranking under the exponential model
2,012
http://arxiv.org/pdf/1206.5265v1
Title Consensus ranking exponential model Summary analyze generalized Mallows model popular exponential model ranking Estimating central consensus ranking data NPhard obtain following new result 1 show search method estimate central ranking pi0 model parameter theta exactly search n worst case tractable true distributi...
[-0.012800532393157482, -0.007486764807254076, 0.0025979948695749044, -0.03712635487318039, 0.005888727493584156, -0.010143857449293137, -0.019638866186141968, -0.03992431238293648, -0.05689305439591408, -0.050429776310920715, 0.010551545768976212, -0.004515605978667736, 0.011392109096050262, 0.02039354480803013, -0.01...
2,068
2,068
['Charles Sutton', 'Andrew McCallum']
1206.5291v1
Belief propagation and its variants are popular methods for approximate inference, but their running time and even their convergence depend greatly on the schedule used to send the messages. Recently, dynamic update schedules have been shown to converge much faster on hard networks than static schedules, namely the res...
Improved Dynamic Schedules for Belief Propagation
2,012
http://arxiv.org/pdf/1206.5291v1
Title Improved Dynamic Schedules Belief Propagation Summary Belief propagation variant popular method approximate inference running time even convergence depend greatly schedule used send message Recently dynamic update schedule shown converge much faster hard network static schedule namely residual BP schedule Elidan ...
[-0.04018958657979965, 0.049813032150268555, -0.006502141710370779, 0.0052742669358849525, -0.06654002517461777, -0.04669348523020744, -0.007517792750149965, -0.03089580498635769, -0.034723296761512756, -0.02761940099298954, 0.08158739656209946, 0.006321096792817116, -0.01582762598991394, 0.06419410556554794, -0.038462...
2,069
2,069
['Yoram Bachrach', 'Thore Graepel', 'Tom Minka', 'John Guiver']
1206.6386v1
We propose a new probabilistic graphical model that jointly models the difficulties of questions, the abilities of participants and the correct answers to questions in aptitude testing and crowdsourcing settings. We devise an active learning/adaptive testing scheme based on a greedy minimization of expected model entro...
How To Grade a Test Without Knowing the Answers --- A Bayesian Graphical Model for Adaptive Crowdsourcing and Aptitude Testing
2,012
http://arxiv.org/pdf/1206.6386v1
Title Grade Test Without Knowing Answers Bayesian Graphical Model Adaptive Crowdsourcing Aptitude Testing Summary propose new probabilistic graphical model jointly model difficulty question ability participant correct answer question aptitude testing crowdsourcing setting devise active learningadaptive testing scheme b...
[0.03895771503448486, 0.027394037693738937, -0.04047328606247902, 0.020326748490333557, -0.01933411881327629, -0.010498049668967724, 0.003794716438278556, 0.022761430591344833, 0.013361758552491665, -0.0040906453505158424, 0.05624779313802719, 0.017590133473277092, 0.0004087270062882453, 0.017273826524615288, 0.0102360...
2,070
2,070
['Jesse Davis', 'Vitor Santos Costa', 'Peggy Peissig', 'Michael Caldwell', 'Elizabeth Berg', 'David Page']
1206.6399v1
Learning from electronic medical records (EMR) is challenging due to their relational nature and the uncertain dependence between a patient's past and future health status. Statistical relational learning is a natural fit for analyzing EMRs but is less adept at handling their inherent latent structure, such as connecti...
Demand-Driven Clustering in Relational Domains for Predicting Adverse Drug Events
2,012
http://arxiv.org/pdf/1206.6399v1
Title DemandDriven Clustering Relational Domains Predicting Adverse Drug Events Summary Learning electronic medical record EMR challenging due relational nature uncertain dependence patient past future health status Statistical relational learning natural fit analyzing EMRs le adept handling inherent latent structure c...
[0.015047644264996052, 0.0331401564180851, -0.020464535802602768, -0.026753058657050133, 0.020620664581656456, 0.01917680911719799, -0.003713468788191676, 0.02898753061890602, 0.06859450042247772, -0.011191675439476967, 0.024885453283786774, -0.01737702079117298, -0.02034551277756691, 0.04123033583164215, -0.0031242575...
2,071
2,071
['Roy Fox', 'Naftali Tishby']
1206.6405v1
In Passive POMDPs actions do not affect the world state, but still incur costs. When the agent is bounded by information-processing constraints, it can only keep an approximation of the belief. We present a variational principle for the problem of maintaining the information which is most useful for minimizing the cost...
Bounded Planning in Passive POMDPs
2,012
http://arxiv.org/pdf/1206.6405v1
Title Bounded Planning Passive POMDPs Summary Passive POMDPs action affect world state still incur cost agent bounded informationprocessing constraint keep approximation belief present variational principle problem maintaining information useful minimizing cost introduce efficient simple algorithm finding optimum Autho...
[0.0005232300027273595, 0.07107508182525635, -0.0029118326492607594, -0.03031974285840988, -0.03622957691550255, -0.05987828969955444, -0.03612849488854408, 0.009806076996028423, -0.02121003530919552, -0.018403448164463043, 0.0689646452665329, 0.05900009348988533, -0.020579461008310318, 0.077720507979393, -0.0126318708...
2,072
2,072
['Takaki Makino', 'Johane Takeuchi']
1206.6484v1
We consider apprenticeship learning, i.e., having an agent learn a task by observing an expert demonstrating the task in a partially observable environment when the model of the environment is uncertain. This setting is useful in applications where the explicit modeling of the environment is difficult, such as a dialog...
Apprenticeship Learning for Model Parameters of Partially Observable Environments
2,012
http://arxiv.org/pdf/1206.6484v1
Title Apprenticeship Learning Model Parameters Partially Observable Environments Summary consider apprenticeship learning ie agent learn task observing expert demonstrating task partially observable environment model environment uncertain setting useful application explicit modeling environment difficult dialogue syste...
[0.017660528421401978, 0.03783974424004555, 0.025048546493053436, -0.021631993353366852, -0.02032013237476349, -0.0010384846245869994, 0.01563578099012375, 0.023083779960870743, 0.01029928121715784, -0.02693767286837101, 0.022142786532640457, 0.039510831236839294, -0.047637321054935455, 0.04460849612951279, 0.011416876...
2,073
2,073
['Avi Pfeffer']
1206.6846v1
One approach to monitoring a dynamic system relies on decomposition of the system into weakly interacting subsystems. An earlier paper introduced a notion of weak interaction called separability, and showed that it leads to exact propagation of marginals for prediction. This paper addresses two questions left open by t...
Approximate Separability for Weak Interaction in Dynamic Systems
2,012
http://arxiv.org/pdf/1206.6846v1
Title Approximate Separability Weak Interaction Dynamic Systems Summary One approach monitoring dynamic system relies decomposition system weakly interacting subsystem earlier paper introduced notion weak interaction called separability showed lead exact propagation marginals prediction paper address two question left ...
[-0.04331379011273384, 0.031162461265921593, -0.013738010078668594, 0.03769551217556, -0.0006120861507952213, -0.01829351671040058, 0.012994579039514065, 0.016228143125772476, 0.006418230477720499, 0.03202848881483078, 0.009688075631856918, -0.015597258694469929, -0.024493789300322533, 0.0728551521897316, -0.0210857298...
2,074
2,074
['Vikash Mansinghka', 'Charles Kemp', 'Thomas Griffiths', 'Joshua Tenenbaum']
1206.6852v1
Traditional approaches to Bayes net structure learning typically assume little regularity in graph structure other than sparseness. However, in many cases, we expect more systematicity: variables in real-world systems often group into classes that predict the kinds of probabilistic dependencies they participate in. Her...
Structured Priors for Structure Learning
2,012
http://arxiv.org/pdf/1206.6852v1
Title Structured Priors Structure Learning Summary Traditional approach Bayes net structure learning typically assume little regularity graph structure sparseness However many case expect systematicity variable realworld system often group class predict kind probabilistic dependency participate capture form prior knowl...
[-0.021279804408550262, 0.0687485784292221, 0.001162343891337514, 0.008381414227187634, 0.00936175137758255, 0.001467838534153998, -0.006710390094667673, 0.003440106986090541, -0.04763364419341087, -2.526411844883114e-05, 0.015947775915265083, -0.008826366625726223, 0.01039489172399044, 0.06735170632600784, 0.047891695...
2,075
2,075
['Sameer Singh', 'Thore Graepel']
1212.0967v1
Instead of requiring a domain expert to specify the probabilistic dependencies of the data, in this work we present an approach that uses the relational DB schema to automatically construct a Bayesian graphical model for a database. This resulting model contains customized distributions for columns, latent variables th...
Compiling Relational Database Schemata into Probabilistic Graphical Models
2,012
http://arxiv.org/pdf/1212.0967v1
Title Compiling Relational Database Schemata Probabilistic Graphical Models Summary Instead requiring domain expert specify probabilistic dependency data work present approach us relational DB schema automatically construct Bayesian graphical model database resulting model contains customized distribution column latent...
[0.014432472176849842, 0.04546206071972847, -0.029537903144955635, -0.024930886924266815, -0.02538476511836052, -0.010074903257191181, 0.005165162496268749, 0.003639871720224619, 0.008448323234915733, -0.01457611471414566, 0.05056304112076759, 0.0011653057299554348, 0.010844264179468155, 0.1135866791009903, 0.009908001...
2,076
2,076
['J. E. Smith', 'P. Caleb-Solly', 'M. A. Tahir', 'D. Sannen', 'H. van-Brussel']
1212.1100v1
The accuracy of machine learning systems is a widely studied research topic. Established techniques such as cross-validation predict the accuracy on unseen data of the classifier produced by applying a given learning method to a given training data set. However, they do not predict whether incurring the cost of obtaini...
Making Early Predictions of the Accuracy of Machine Learning Applications
2,012
http://arxiv.org/pdf/1212.1100v1
Title Making Early Predictions Accuracy Machine Learning Applications Summary accuracy machine learning system widely studied research topic Established technique crossvalidation predict accuracy unseen data classifier produced applying given learning method given training data set However predict whether incurring cos...
[0.005124268587678671, 0.0031944497022777796, -0.03592777997255325, -0.029026586562395096, 0.004230910912156105, -0.01283400971442461, 0.0392209030687809, 0.0011758188484236598, 0.004601708613336086, -0.0029514299239963293, 0.022366909310221672, 0.01232985407114029, 0.0341201089322567, 0.055009160190820694, -0.00224649...
2,077
2,077
['David Maxwell Chickering', 'David Heckerman']
1302.3567v2
We discuss Bayesian methods for learning Bayesian networks when data sets are incomplete. In particular, we examine asymptotic approximations for the marginal likelihood of incomplete data given a Bayesian network. We consider the Laplace approximation and the less accurate but more efficient BIC/MDL approximation. We ...
Efficient Approximations for the Marginal Likelihood of Incomplete Data Given a Bayesian Network
2,013
http://arxiv.org/pdf/1302.3567v2
Title Efficient Approximations Marginal Likelihood Incomplete Data Given Bayesian Network Summary discus Bayesian method learning Bayesian network data set incomplete particular examine asymptotic approximation marginal likelihood incomplete data given Bayesian network consider Laplace approximation le accurate efficie...
[-0.05530932545661926, -0.008555022068321705, -0.012237011454999447, 0.014796613715589046, -0.02330068126320839, -0.04140667989850044, 0.01419099047780037, -0.02040012553334236, -0.017738278955221176, 0.038557425141334534, 0.0631520226597786, -0.005930772051215172, 0.023651497438549995, 0.011331220157444477, -0.0231454...
2,078
2,078
['Dan Geiger', 'David Heckerman', 'Christopher Meek']
1302.3580v2
We extend the Bayesian Information Criterion (BIC), an asymptotic approximation for the marginal likelihood, to Bayesian networks with hidden variables. This approximation can be used to select models given large samples of data. The standard BIC as well as our extension punishes the complexity of a model according to ...
Asymptotic Model Selection for Directed Networks with Hidden Variables
2,013
http://arxiv.org/pdf/1302.3580v2
Title Asymptotic Model Selection Directed Networks Hidden Variables Summary extend Bayesian Information Criterion BIC asymptotic approximation marginal likelihood Bayesian network hidden variable approximation used select model given large sample data standard BIC well extension punishes complexity model according dime...
[-0.05903637781739235, 0.02550484612584114, -0.01943427324295044, -0.011078529991209507, 0.0019408465595915914, 0.0022101118229329586, 0.05782820284366608, 0.01247506681829691, 0.01597348228096962, 0.021665949374437332, 0.04662551358342171, 0.05153018236160278, 0.008386794477701187, 0.0787314921617508, 0.01366748195141...
2,079
2,079
['Hado van Hasselt']
1302.7175v2
We investigate the accuracy of the two most common estimators for the maximum expected value of a general set of random variables: a generalization of the maximum sample average, and cross validation. No unbiased estimator exists and we show that it is non-trivial to select a good estimator without knowledge about the ...
Estimating the Maximum Expected Value: An Analysis of (Nested) Cross Validation and the Maximum Sample Average
2,013
http://arxiv.org/pdf/1302.7175v2
Title Estimating Maximum Expected Value Analysis Nested Cross Validation Maximum Sample Average Summary investigate accuracy two common estimator maximum expected value general set random variable generalization maximum sample average cross validation unbiased estimator exists show nontrivial select good estimator with...
[-0.037611596286296844, -0.03296448290348053, -0.0270608589053154, -0.008480421267449856, 0.017933005467057228, 6.520982424262911e-05, 0.04055718705058098, -0.011817376129329205, 0.03460663557052612, 0.015888717025518417, 0.014674766920506954, 0.04608674719929695, 0.04094269499182701, 0.047166649252176285, 0.0286987815...
2,080
2,080
['Somayeh Danafar', 'Paola M. V. Rancoita', 'Tobias Glasmachers', 'Kevin Whittingstall', 'Juergen Schmidhuber']
1305.0423v1
Do two data samples come from different distributions? Recent studies of this fundamental problem focused on embedding probability distributions into sufficiently rich characteristic Reproducing Kernel Hilbert Spaces (RKHSs), to compare distributions by the distance between their embeddings. We show that Regularized Ma...
Testing Hypotheses by Regularized Maximum Mean Discrepancy
2,013
http://arxiv.org/pdf/1305.0423v1
Title Testing Hypotheses Regularized Maximum Mean Discrepancy Summary two data sample come different distribution Recent study fundamental problem focused embedding probability distribution sufficiently rich characteristic Reproducing Kernel Hilbert Spaces RKHSs compare distribution distance embeddings show Regularized...
[-0.03356059268116951, -0.015980107709765434, -0.03890058025717735, 0.0005031739128753543, -0.01752518303692341, 0.006541328504681587, -0.01033998467028141, 0.03200575336813927, 0.02324211224913597, 0.05152469500899315, -0.0009081657626666129, -0.015500600449740887, 0.062250811606645584, 0.011886447668075562, 0.0583701...
2,081
2,081
['Aaron Karper']
1402.2300v1
The amount of information in the form of features and variables avail- able to machine learning algorithms is ever increasing. This can lead to classifiers that are prone to overfitting in high dimensions, high di- mensional models do not lend themselves to interpretable results, and the CPU and memory resources necess...
Feature and Variable Selection in Classification
2,014
http://arxiv.org/pdf/1402.2300v1
Title Feature Variable Selection Classification Summary amount information form feature variable avail able machine learning algorithm ever increasing lead classifier prone overfitting high dimension high di mensional model lend interpretable result CPU memory resource necessary run highdimensional datasets severly lim...
[-0.01503723580390215, 0.03310861065983772, -0.05072430521249771, 0.01964566856622696, 0.004890934564173222, 0.00293405307456851, 0.05864860862493515, 0.028686795383691788, 0.005625756457448006, 0.005444481503218412, 0.06475509703159332, 0.04742700979113579, 0.00784587673842907, 0.05869772285223007, -0.0156638063490390...
2,082
2,082
['Lihong Li', 'Shunbao Chen', 'Jim Kleban', 'Ankur Gupta']
1403.1891v2
Optimizing an interactive system against a predefined online metric is particularly challenging, when the metric is computed from user feedback such as clicks and payments. The key challenge is the counterfactual nature: in the case of Web search, any change to a component of the search engine may result in a different...
Counterfactual Estimation and Optimization of Click Metrics for Search Engines
2,014
http://arxiv.org/pdf/1403.1891v2
Title Counterfactual Estimation Optimization Click Metrics Search Engines Summary Optimizing interactive system predefined online metric particularly challenging metric computed user feedback click payment key challenge counterfactual nature case Web search change component search engine may result different search res...
[0.026970619335770607, 0.057925522327423096, -0.038612689822912216, -0.011467079631984234, -0.02481050044298172, -0.03715641051530838, 0.03110860101878643, 0.03522871434688568, 0.006067683920264244, -0.06256449967622757, -0.03252210095524788, -0.0003582848294172436, -0.012323129922151566, 0.09273400902748108, 0.0200419...
2,083
2,083
['Zhana Bao']
1405.1027v1
Finding rare information hidden in a huge amount of data from the Internet is a necessary but complex issue. Many researchers have studied this issue and have found effective methods to detect anomaly data in low dimensional space. However, as the dimension increases, most of these existing methods perform poorly in de...
K-NS: Section-Based Outlier Detection in High Dimensional Space
2,014
http://arxiv.org/pdf/1405.1027v1
Title KNS SectionBased Outlier Detection High Dimensional Space Summary Finding rare information hidden huge amount data Internet necessary complex issue Many researcher studied issue found effective method detect anomaly data low dimensional space However dimension increase existing method perform poorly detecting out...
[0.009748214855790138, -0.018245546147227287, -0.03769798204302788, 0.041060637682676315, -0.018476534634828568, 0.01318865641951561, 0.028490252792835236, 0.06297449767589569, -0.024636516347527504, 0.041691116988658905, 0.0538514107465744, 0.04608277231454849, -0.007786982227116823, 0.056568119674921036, -0.031496793...
2,084
2,084
['Ji Zhao', 'Deyu Meng']
1405.2664v2
The maximum mean discrepancy (MMD) is a recently proposed test statistic for two-sample test. Its quadratic time complexity, however, greatly hampers its availability to large-scale applications. To accelerate the MMD calculation, in this study we propose an efficient method called FastMMD. The core idea of FastMMD is ...
FastMMD: Ensemble of Circular Discrepancy for Efficient Two-Sample Test
2,014
http://arxiv.org/pdf/1405.2664v2
Title FastMMD Ensemble Circular Discrepancy Efficient TwoSample Test Summary maximum mean discrepancy MMD recently proposed test statistic twosample test quadratic time complexity however greatly hamper availability largescale application accelerate MMD calculation study propose efficient method called FastMMD core ide...
[-0.05340200290083885, -0.004506088327616453, -0.05934504047036171, 0.05318877100944519, -0.008252672851085663, -0.032670196145772934, 0.06480226665735245, -0.02573218196630478, -0.024504218250513077, 0.027538031339645386, 0.014620550908148289, -0.018866494297981262, 0.05498983711004257, -0.029056239873170853, -0.00863...
2,085
2,085
['Hongwei Li', 'Qiang Liu']
1502.00725v1
Crowdsourcing provides a popular paradigm for data collection at scale. We study the problem of selecting subsets of workers from a given worker pool to maximize the accuracy under a budget constraint. One natural question is whether we should hire as many workers as the budget allows, or restrict on a small number of ...
Cheaper and Better: Selecting Good Workers for Crowdsourcing
2,015
http://arxiv.org/pdf/1502.00725v1
Title Cheaper Better Selecting Good Workers Crowdsourcing Summary Crowdsourcing provides popular paradigm data collection scale study problem selecting subset worker given worker pool maximize accuracy budget constraint One natural question whether hire many worker budget allows restrict small number topquality worker ...
[0.016991762444376945, 0.06856611371040344, -0.03477822244167328, -0.03048105351626873, -0.029157651588320732, -0.059212084859609604, 0.0466863252222538, 0.04394686967134476, -0.015075954608619213, -0.04038483649492264, 0.04378972575068474, 0.04919582977890968, -0.003046088619157672, -0.0033984200563281775, 0.022043546...
2,086
2,086
['Shuai Li', 'Alexandros Karatzoglou', 'Claudio Gentile']
1502.03473v7
Classical collaborative filtering, and content-based filtering methods try to learn a static recommendation model given training data. These approaches are far from ideal in highly dynamic recommendation domains such as news recommendation and computational advertisement, where the set of items and users is very fluid....
Collaborative Filtering Bandits
2,015
http://arxiv.org/pdf/1502.03473v7
Title Collaborative Filtering Bandits Summary Classical collaborative filtering contentbased filtering method try learn static recommendation model given training data approach far ideal highly dynamic recommendation domain news recommendation computational advertisement set item user fluid work investigate adaptive cl...
[0.043313391506671906, 0.004283931106328964, -0.01708991266787052, -0.019534491002559662, 0.0005997842526994646, -0.011954969726502895, 0.042941562831401825, 0.021548865363001823, 0.01073745172470808, 0.006185795646160841, -0.053907983005046844, 0.00370569690130651, -0.02590211294591427, 0.03241090476512909, -0.0068773...
2,087
2,087
['Jacob Abernethy', 'Yiling Chen', 'Chien-Ju Ho', 'Bo Waggoner']
1502.05774v2
We design mechanisms for online procurement of data held by strategic agents for machine learning tasks. The challenge is to use past data to actively price future data and give learning guarantees even when an agent's cost for revealing her data may depend arbitrarily on the data itself. We achieve this goal by showin...
Low-Cost Learning via Active Data Procurement
2,015
http://arxiv.org/pdf/1502.05774v2
Title LowCost Learning via Active Data Procurement Summary design mechanism online procurement data held strategic agent machine learning task challenge use past data actively price future data give learning guarantee even agent cost revealing data may depend arbitrarily data achieve goal showing convert large class no...
[-0.023585449904203415, 0.08519540727138519, -0.017772499471902847, -0.05586612597107887, -0.006209905259311199, -0.03100406937301159, 0.06546367704868317, 0.021294724196195602, -0.022671116515994072, -0.01102172490209341, 0.015225880779325962, 0.04702921584248543, -0.017882371321320534, 0.13014262914657593, -0.0144842...
2,088
2,088
['Md Amran Siddiqui', 'Alan Fern', 'Thomas G. Dietterich', 'Weng-Keen Wong']
1503.00038v1
In many applications, an anomaly detection system presents the most anomalous data instance to a human analyst, who then must determine whether the instance is truly of interest (e.g. a threat in a security setting). Unfortunately, most anomaly detectors provide no explanation about why an instance was considered anoma...
Sequential Feature Explanations for Anomaly Detection
2,015
http://arxiv.org/pdf/1503.00038v1
Title Sequential Feature Explanations Anomaly Detection Summary many application anomaly detection system present anomalous data instance human analyst must determine whether instance truly interest eg threat security setting Unfortunately anomaly detector provide explanation instance considered anomalous leaving analy...
[0.03856004774570465, -0.007319788448512554, -0.012067083269357681, 0.03410965949296951, -0.04685444384813309, -0.01119364146143198, 0.04849904403090477, 0.008157940581440926, -0.02130657434463501, -0.02270738035440445, 0.1022520586848259, -0.0008323031943291426, -0.011948407627642155, 0.12237460911273956, -0.021566396...
2,089
2,089
['Andrew Emmott', 'Shubhomoy Das', 'Thomas Dietterich', 'Alan Fern', 'Weng-Keen Wong']
1503.01158v2
This article provides a thorough meta-analysis of the anomaly detection problem. To accomplish this we first identify approaches to benchmarking anomaly detection algorithms across the literature and produce a large corpus of anomaly detection benchmarks that vary in their construction across several dimensions we deem...
A Meta-Analysis of the Anomaly Detection Problem
2,015
http://arxiv.org/pdf/1503.01158v2
Title MetaAnalysis Anomaly Detection Problem Summary article provides thorough metaanalysis anomaly detection problem accomplish first identify approach benchmarking anomaly detection algorithm across literature produce large corpus anomaly detection benchmark vary construction across several dimension deem important r...
[0.06328872591257095, -0.020624376833438873, -0.032865215092897415, 0.01918272115290165, -0.03468652069568634, 0.00162624625954777, 0.04972092807292938, 0.01620400696992874, -0.04520917311310768, -0.02940533310174942, 0.048684753477573395, 0.0182701013982296, -0.003590296721085906, 0.05678805708885193, -0.0417051911354...
2,090
2,090
['Qingming Tang', 'Siqi Sun', 'Jinbo Xu']
1503.02129v3
Learning the network structure underlying data is an important problem in machine learning. This paper introduces a novel prior to study the inference of scale-free networks, which are widely used to model social and biological networks. The prior not only favors a desirable global node degree distribution, but also ta...
Learning Scale-Free Networks by Dynamic Node-Specific Degree Prior
2,015
http://arxiv.org/pdf/1503.02129v3
Title Learning ScaleFree Networks Dynamic NodeSpecific Degree Prior Summary Learning network structure underlying data important problem machine learning paper introduces novel prior study inference scalefree network widely used model social biological network prior favor desirable global node degree distribution also ...
[-0.017744367942214012, -0.00891782809048891, -0.03865165263414383, -0.018091997131705284, -0.009269659407436848, -0.017756953835487366, 0.03190317004919052, 0.006770459935069084, 0.05484851449728012, 0.024441951885819435, 0.01901293732225895, 0.04002838954329491, -0.00866029504686594, 0.10840190947055817, 0.0022172452...
2,091
2,091
['Qinqing Zheng', 'Ryota Tomioka']
1503.05479v2
We consider the problem of recovering a low-rank tensor from its noisy observation. Previous work has shown a recovery guarantee with signal to noise ratio $O(n^{\lceil K/2 \rceil /2})$ for recovering a $K$th order rank one tensor of size $n\times \cdots \times n$ by recursive unfolding. In this paper, we first improve...
Interpolating Convex and Non-Convex Tensor Decompositions via the Subspace Norm
2,015
http://arxiv.org/pdf/1503.05479v2
Title Interpolating Convex NonConvex Tensor Decompositions via Subspace Norm Summary consider problem recovering lowrank tensor noisy observation Previous work shown recovery guarantee signal noise ratio Onlceil K2 rceil 2 recovering Kth order rank one tensor size ntimes cdots time n recursive unfolding paper first imp...
[-0.025488197803497314, 0.07685485482215881, 0.005058417096734047, 0.03756030648946762, -0.0010564718395471573, 0.02929343655705452, 0.005151380784809589, 0.07240483164787292, -0.030262667685747147, 0.013362454250454903, 0.004248038865625858, -0.04480091854929924, -0.0028084793593734503, -0.02914496324956417, 0.0015379...
2,092
2,092
['Jinye Zhang', 'Zhijian Ou']
1503.06239v1
Existing MAP inference algorithms for determinantal point processes (DPPs) need to calculate determinants or conduct eigenvalue decomposition generally at the scale of the full kernel, which presents a great challenge for real-world applications. In this paper, we introduce a class of DPPs, called BwDPPs, that are char...
Block-Wise MAP Inference for Determinantal Point Processes with Application to Change-Point Detection
2,015
http://arxiv.org/pdf/1503.06239v1
Title BlockWise MAP Inference Determinantal Point Processes Application ChangePoint Detection Summary Existing MAP inference algorithm determinantal point process DPPs need calculate determinant conduct eigenvalue decomposition generally scale full kernel present great challenge realworld application paper introduce cl...
[-0.07240423560142517, 0.000522443326190114, -0.04835313931107521, -0.007268576882779598, -0.015433134511113167, -0.00039502704748883843, 0.03038032166659832, 0.024285001680254936, 0.04154610261321068, 0.02257649227976799, 0.07686968147754669, 0.05603582784533501, 0.029130972921848297, 0.01931862160563469, -0.006954280...
2,093
2,093
['Taylor Killian', 'George Konidaris', 'Finale Doshi-Velez']
1612.00475v1
Due to physiological variation, patients diagnosed with the same condition may exhibit divergent, but related, responses to the same treatments. Hidden Parameter Markov Decision Processes (HiP-MDPs) tackle this transfer-learning problem by embedding these tasks into a low-dimensional space. However, the original formul...
Transfer Learning Across Patient Variations with Hidden Parameter Markov Decision Processes
2,016
http://arxiv.org/pdf/1612.00475v1
Title Transfer Learning Across Patient Variations Hidden Parameter Markov Decision Processes Summary Due physiological variation patient diagnosed condition may exhibit divergent related response treatment Hidden Parameter Markov Decision Processes HiPMDPs tackle transferlearning problem embedding task lowdimensional s...
[0.014823011122643948, 0.03425592929124832, -0.01384636852890253, -0.05035480856895447, 0.02991417795419693, 0.01956239901483059, 0.019398363307118416, 0.0203845277428627, -0.012039663270115852, 0.013737629167735577, 0.006257784087210894, -0.011194316670298576, 0.013564607128500938, 0.033341530710458755, -0.00585546065...
2,094
2,094
['James Kirkpatrick', 'Razvan Pascanu', 'Neil Rabinowitz', 'Joel Veness', 'Guillaume Desjardins', 'Andrei A. Rusu', 'Kieran Milan', 'John Quan', 'Tiago Ramalho', 'Agnieszka Grabska-Barwinska', 'Demis Hassabis', 'Claudia Clopath', 'Dharshan Kumaran', 'Raia Hadsell']
1612.00796v2
The ability to learn tasks in a sequential fashion is crucial to the development of artificial intelligence. Neural networks are not, in general, capable of this and it has been widely thought that catastrophic forgetting is an inevitable feature of connectionist models. We show that it is possible to overcome this lim...
Overcoming catastrophic forgetting in neural networks
2,016
http://arxiv.org/pdf/1612.00796v2
Title Overcoming catastrophic forgetting neural network Summary ability learn task sequential fashion crucial development artificial intelligence Neural network general capable widely thought catastrophic forgetting inevitable feature connectionist model show possible overcome limitation train network maintain expertis...
[-0.013627736829221249, 0.06784623861312866, -0.0245034322142601, 0.0006046528578735888, 0.02747356705367565, 0.017595097422599792, -0.0224929042160511, -0.018583137542009354, -0.01327467244118452, 0.030320804566144943, 0.013708880171179771, 0.010989083908498287, -0.023444807156920433, 0.02875937707722187, 0.0185731071...
2,095
2,095
['Peter Karkus', 'Andras Kupcsik', 'David Hsu', 'Wee Sun Lee']
1612.01746v1
Scarce data is a major challenge to scaling robot learning to truly complex tasks, as we need to generalize locally learned policies over different "contexts". Bayesian optimization approaches to contextual policy search (CPS) offer data-efficient policy learning that generalize over a context space. We propose to impr...
Factored Contextual Policy Search with Bayesian Optimization
2,016
http://arxiv.org/pdf/1612.01746v1
Title Factored Contextual Policy Search Bayesian Optimization Summary Scarce data major challenge scaling robot learning truly complex task need generalize locally learned policy different context Bayesian optimization approach contextual policy search CPS offer dataefficient policy learning generalize context space pr...
[-0.021594161167740822, -0.0009851943468675017, 0.03450219705700874, -0.0039422414265573025, 0.005831927061080933, -0.015931300818920135, 0.013361268676817417, 0.0013885862426832318, -0.051722656935453415, -0.017539044842123985, -0.01869313232600689, 0.00790545903146267, 0.0035008389968425035, 0.03543585538864136, -0.0...
2,096
2,096
['Ting Chen', 'Yizhou Sun']
1612.02814v2
In this paper, we study the problem of author identification under double-blind review setting, which is to identify potential authors given information of an anonymized paper. Different from existing approaches that rely heavily on feature engineering, we propose to use network embedding approach to address the proble...
Task-Guided and Path-Augmented Heterogeneous Network Embedding for Author Identification
2,016
http://arxiv.org/pdf/1612.02814v2
Title TaskGuided PathAugmented Heterogeneous Network Embedding Author Identification Summary paper study problem author identification doubleblind review setting identify potential author given information anonymized paper Different existing approach rely heavily feature engineering propose use network embedding approa...
[0.006971737835556269, -0.022771546617150307, -0.023734131827950478, 0.02401071973145008, 0.017898106947541237, -0.018154677003622055, 0.08214766532182693, -0.03914443403482437, 0.040595944970846176, -0.021216291934251785, -0.01025584526360035, 0.08158876746892929, -0.030711006373167038, 0.04144924879074097, 0.03559555...
2,097
2,097
['Vivek Veeriah', 'Shangtong Zhang', 'Richard S. Sutton']
1612.02879v2
Representations are fundamental to artificial intelligence. The performance of a learning system depends on the type of representation used for representing the data. Typically, these representations are hand-engineered using domain knowledge. More recently, the trend is to learn these representations through stochasti...
Learning Representations by Stochastic Meta-Gradient Descent in Neural Networks
2,016
http://arxiv.org/pdf/1612.02879v2
Title Learning Representations Stochastic MetaGradient Descent Neural Networks Summary Representations fundamental artificial intelligence performance learning system depends type representation used representing data Typically representation handengineered using domain knowledge recently trend learn representation sto...
[-0.02671397663652897, -0.007372573018074036, -0.012821019627153873, -0.01619107834994793, 0.0076756603084504604, -0.030694853514432907, 0.033791832625865936, -0.009645365178585052, -0.03539589047431946, -0.008695905096828938, -0.024991288781166077, 0.042916152626276016, 0.008951188065111637, 0.06423050165176392, 0.010...
2,098
2,098
['Hanie Sedghi', 'Ashish Sabharwal']
1612.03871v2
Given a knowledge base (KB) rich in facts about common nouns or generics, such as "all trees produce oxygen" or "some animals live in forests", we consider the problem of deriving additional such facts at a high precision. While this problem has received much attention for named entity KBs such as Freebase, little emph...
Knowledge Completion for Generics using Guided Tensor Factorization
2,016
http://arxiv.org/pdf/1612.03871v2
Title Knowledge Completion Generics using Guided Tensor Factorization Summary Given knowledge base KB rich fact common noun generic tree produce oxygen animal live forest consider problem deriving additional fact high precision problem received much attention named entity KBs Freebase little emphasis placed generic des...
[0.024226149544119835, 0.04627448320388794, -0.0032957098446786404, 0.016702936962246895, 0.0232491884380579, 0.0330202616751194, -0.018531620502471924, 0.046567704528570175, -0.03186938911676407, -0.06299469619989395, 0.015076507814228535, 0.022237233817577362, 0.015674546360969543, 0.028310386463999748, -0.0095683727...
2,099
2,099
['Marco F. Cusumano-Towner', 'Vikash K. Mansinghka']
1612.04759v2
This paper introduces the probabilistic module interface, which allows encapsulation of complex probabilistic models with latent variables alongside custom stochastic approximate inference machinery, and provides a platform-agnostic abstraction barrier separating the model internals from the host probabilistic inferenc...
Encapsulating models and approximate inference programs in probabilistic modules
2,016
http://arxiv.org/pdf/1612.04759v2
Title Encapsulating model approximate inference program probabilistic module Summary paper introduces probabilistic module interface allows encapsulation complex probabilistic model latent variable alongside custom stochastic approximate inference machinery provides platformagnostic abstraction barrier separating model...
[-0.026385894045233727, 0.06399696320295334, -0.0004334723053034395, -0.009615359827876091, -0.031692229211330414, -0.033143073320388794, 0.030327208340168, -0.06841228157281876, -0.0075864423997700214, 0.025383543223142624, 0.02685406617820263, 0.02124253660440445, -0.004541034810245037, 0.10739387571811676, 0.0463104...