title
string
paper_url
string
authors
list
type
string
primary_area
string
abstract
large_string
keywords
list
TL;DR
large_string
submission_number
int64
arxiv_id
string
arxiv_id_source
string
embedding
list
Stochastic Optimization with Importance Sampling for Regularized Loss Minimization
https://proceedings.mlr.press/v37/zhaoa15.html
[ "Peilin Zhao", "Tong Zhang" ]
null
null
Uniform sampling of training data has been commonly used in traditional stochastic optimization algorithms such as Proximal Stochastic Mirror Descent (prox-SMD) and Proximal Stochastic Dual Coordinate Ascent (prox-SDCA). Although uniform sampling can guarantee that the sampled stochastic quantity is an unbiased estimat...
[]
null
1
null
null
[ -0.028318965807557106, -0.026734089478850365, 0.019209325313568115, 0.0630248561501503, 0.04426506161689758, 0.04000489413738251, 0.010027599520981312, -0.004918746650218964, -0.01214030385017395, -0.057430200278759, -0.00005432681427919306, -0.020212344825267792, -0.041260067373514175, -0...
Approval Voting and Incentives in Crowdsourcing
https://proceedings.mlr.press/v37/shaha15.html
[ "Nihar Shah", "Dengyong Zhou", "Yuval Peres" ]
null
null
The growing need for labeled training data has made crowdsourcing an important part of machine learning. The quality of crowdsourced labels is, however, adversely affected by three factors: (1) the workers are not experts; (2) the incentives of the workers are not aligned with those of the requesters; and (3) the inter...
[]
null
2
1502.05696
title_snapshot
[ 0.011874196119606495, -0.03152801841497421, -0.014264218509197235, 0.03764692321419716, 0.018492475152015686, 0.01872008666396141, -0.00032475803163833916, -0.0048646447248756886, -0.034701116383075714, -0.021923795342445374, -0.02526642009615898, 0.01796731725335121, -0.03691907227039337, ...
A low variance consistent test of relative dependency
https://proceedings.mlr.press/v37/bounliphone15.html
[ "Wacha Bounliphone", "Arthur Gretton", "Arthur Tenenhaus", "Matthew Blaschko" ]
null
null
We describe a novel non-parametric statistical hypothesis test of relative dependence between a source variable and two candidate target variables. Such a test enables us to determine whether one source variable is significantly more dependent on a first target variable or a second. Dependence is measured via the Hilbe...
[]
null
3
1406.3852
title_snapshot
[ -0.02162117138504982, 0.023915685713291168, -0.02305126190185547, -0.00886557623744011, 0.061364080756902695, 0.031250521540641785, 0.03256375715136528, -0.01693417876958847, -0.023718336597085, -0.03209324926137924, 0.016877412796020508, 0.018902141600847244, -0.05534173175692558, 0.01393...
An Aligned Subtree Kernel for Weighted Graphs
https://proceedings.mlr.press/v37/bai15.html
[ "Lu Bai", "Luca Rossi", "Zhihong Zhang", "Edwin Hancock" ]
null
null
In this paper, we develop a new entropic matching kernel for weighted graphs by aligning depth-based representations. We demonstrate that this kernel can be seen as an \textbfaligned subtree kernel that incorporates explicit subtree correspondences, and thus addresses the drawback of neglecting the relative locations b...
[]
null
4
null
null
[ 0.009152614511549473, -0.03351321816444397, 0.023153511807322502, 0.05977610871195793, 0.0050128852017223835, 0.05955234915018082, 0.01396131794899702, -0.00291979918256402, 0.00797563698142767, -0.054034288972616196, -0.021644284948706627, -0.01463281735777855, -0.06523551791906357, -0.00...
Spectral Clustering via the Power Method - Provably
https://proceedings.mlr.press/v37/boutsidis15.html
[ "Christos Boutsidis", "Prabhanjan Kambadur", "Alex Gittens" ]
null
null
Spectral clustering is one of the most important algorithms in data mining and machine intelligence; however, its computational complexity limits its application to truly large scale data analysis. The computational bottleneck in spectral clustering is computing a few of the top eigenvectors of the (normalized) Laplaci...
[]
null
5
1311.2854
title_snapshot
[ -0.024880090728402138, -0.01918053813278675, 0.0024078928399831057, 0.0336647666990757, 0.05762956663966179, 0.02688758634030819, 0.03454461693763733, -0.004141977522522211, 0.0011258251033723354, -0.04563852772116661, -0.012508914805948734, -0.02944096177816391, -0.05771436542272568, -0.0...
Information Geometry and Minimum Description Length Networks
https://proceedings.mlr.press/v37/suna15.html
[ "Ke Sun", "Jun Wang", "Alexandros Kalousis", "Stephan Marchand-Maillet" ]
null
null
We study parametric unsupervised mixture learning. We measure the loss of intrinsic information from the observations to complex mixture models, and then to simple mixture models. We present a geometric picture, where all these representations are regarded as free points in the space of probability distributions. Based...
[]
null
6
null
null
[ -0.004370131995528936, -0.003339793998748064, -0.01091594435274601, 0.044968847185373306, 0.02026538923382759, 0.029560916125774384, 0.025445925071835518, 0.023866256698966026, -0.02930217608809471, -0.0351150780916214, -0.010684589855372906, 0.028252452611923218, -0.04605213552713394, 0.0...
Efficient Training of LDA on a GPU by Mean-for-Mode Estimation
https://proceedings.mlr.press/v37/tristan15.html
[ "Jean-Baptiste Tristan", "Joseph Tassarotti", "Guy Steele" ]
null
null
We introduce Mean-for-Mode estimation, a variant of an uncollapsed Gibbs sampler that we use to train LDA on a GPU. The algorithm combines benefits of both uncollapsed and collapsed Gibbs samplers. Like a collapsed Gibbs sampler — and unlike an uncollapsed Gibbs sampler — it has good statistical performance, and can us...
[]
null
7
null
null
[ 0.0062541901133954525, -0.01194192748516798, 0.007233529817312956, 0.03672471269965172, 0.02747713029384613, 0.025588948279619217, 0.02658381126821041, 0.03685929253697395, -0.014925003051757812, -0.05460559204220772, -0.028625141829252243, 0.0062574404291808605, -0.08330829441547394, 0.00...
Adaptive Stochastic Alternating Direction Method of Multipliers
https://proceedings.mlr.press/v37/zhaob15.html
[ "Peilin Zhao", "Jinwei Yang", "Tong Zhang", "Ping Li" ]
null
null
The Alternating Direction Method of Multipliers (ADMM) has been studied for years. Traditional ADMM algorithms need to compute, at each iteration, an (empirical) expected loss function on all training examples, resulting in a computational complexity proportional to the number of training examples. To reduce the comple...
[]
null
8
1312.4564
title_snapshot
[ -0.04438110440969467, -0.012912645936012268, 0.01311451941728592, -0.0057127769105136395, 0.019156768918037415, 0.06413222849369049, 0.03923797607421875, -0.012988659553229809, -0.05687037855386734, -0.0363520011305809, -0.025561515241861343, -0.004406827501952648, -0.03426055610179901, -0...
A Lower Bound for the Optimization of Finite Sums
https://proceedings.mlr.press/v37/agarwal15.html
[ "Alekh Agarwal", "Leon Bottou" ]
null
null
This paper presents a lower bound for optimizing a finite sum of n functions, where each function is L-smooth and the sum is μ-strongly convex. We show that no algorithm can reach an error εin minimizing all functions from this class in fewer than Ω(n + \sqrtn(κ-1)\log(1/ε)) iterations, where κ=L/μis a surrogate condit...
[]
null
9
1410.0723
title_snapshot
[ -0.04839242249727249, -0.0006392368231900036, 0.03384550288319588, 0.021054605022072792, 0.0461343452334404, 0.04140862077474594, 0.019710954278707504, -0.01776091940701008, -0.01861061528325081, -0.02613898180425167, -0.007374874781817198, 0.005831785965710878, -0.06947455555200577, -0.00...
Learning Word Representations with Hierarchical Sparse Coding
https://proceedings.mlr.press/v37/yogatama15.html
[ "Dani Yogatama", "Manaal Faruqui", "Chris Dyer", "Noah Smith" ]
null
null
We propose a new method for learning word representations using hierarchical regularization in sparse coding inspired by the linguistic study of word meanings. We show an efficient learning algorithm based on stochastic proximal methods that is significantly faster than previous approaches, making it possible to perfor...
[]
null
10
1406.2035
title_snapshot
[ -0.023175621405243874, 0.001279569580219686, 0.021591830998659134, 0.04106441140174866, 0.03733716160058975, 0.05120890215039253, 0.026867683976888657, 0.01611199788749218, -0.03504985198378563, -0.03721441701054573, -0.0013131361920386553, -0.0066981627605855465, -0.0496351383626461, -0.0...
Learning Transferable Features with Deep Adaptation Networks
https://proceedings.mlr.press/v37/long15.html
[ "Mingsheng Long", "Yue Cao", "Jianmin Wang", "Michael Jordan" ]
null
null
Recent studies reveal that a deep neural network can learn transferable features which generalize well to novel tasks for domain adaptation. However, as deep features eventually transition from general to specific along the network, the feature transferability drops significantly in higher layers with increasing domain...
[]
null
11
1502.02791
title_snapshot
[ -0.01840735599398613, -0.015263248234987259, 0.014912464655935764, 0.03390119597315788, 0.060130033642053604, 0.033098071813583374, 0.027000347152352333, -0.015903418883681297, 0.018316728994250298, -0.037910036742687225, -0.02070626989006996, 0.013044824823737144, -0.05802726000547409, 0....
Robust partially observable Markov decision process
https://proceedings.mlr.press/v37/osogami15.html
[ "Takayuki Osogami" ]
null
null
We seek to find the robust policy that maximizes the expected cumulative reward for the worst case when a partially observable Markov decision process (POMDP) has uncertain parameters whose values are only known to be in a given region. We prove that the robust value function, which represents the expected cumulative r...
[]
null
12
null
null
[ -0.0566207654774189, 0.002464734483510256, -0.014788232743740082, 0.0543217770755291, 0.04595354571938515, 0.036116935312747955, 0.017931882292032242, 0.007878012955188751, -0.01168518140912056, -0.06439971923828125, -0.04536259174346924, -0.02399066649377346, -0.07651606947183609, -0.0202...
On the Relationship between Sum-Product Networks and Bayesian Networks
https://proceedings.mlr.press/v37/zhaoc15.html
[ "Han Zhao", "Mazen Melibari", "Pascal Poupart" ]
null
null
In this paper, we establish some theoretical connections between Sum-Product Networks (SPNs) and Bayesian Networks (BNs). We prove that every SPN can be converted into a BN in linear time and space in terms of the network size. The key insight is to use Algebraic Decision Diagrams (ADDs) to compactly represent the loca...
[]
null
13
1501.01239
title_snapshot
[ -0.02375628799200058, 0.037904027849435806, 0.004514158237725496, 0.026313437148928642, 0.05720498412847519, 0.024488983675837517, 0.023381615057587624, -0.003822057042270899, -0.020592542365193367, -0.033200524747371674, -0.0035741967149078846, 0.02692701667547226, -0.058661554008722305, ...
Learning from Corrupted Binary Labels via Class-Probability Estimation
https://proceedings.mlr.press/v37/menon15.html
[ "Aditya Menon", "Brendan Van Rooyen", "Cheng Soon Ong", "Bob Williamson" ]
null
null
Many supervised learning problems involve learning from samples whose labels are corrupted in some way. For example, each sample may have some constant probability of being incorrectly labelled (learning with label noise), or one may have a pool of unlabelled samples in lieu of negative samples (learning from positive ...
[]
null
14
null
null
[ 0.023735182359814644, 0.0062027545645833015, -0.06137178838253021, 0.07662343233823776, 0.04028855264186859, 0.030434992164373398, -0.003902341704815626, -0.007560905069112778, -0.01066221296787262, -0.018741002306342125, -0.011851254850625992, 0.03460146114230156, -0.07900000363588333, 0....
An Explicit Sampling Dependent Spectral Error Bound for Column Subset Selection
https://proceedings.mlr.press/v37/yanga15.html
[ "Tianbao Yang", "Lijun Zhang", "Rong Jin", "Shenghuo Zhu" ]
null
null
In this paper, we consider the problem of column subset selection. We present a novel analysis of the spectral norm reconstruction for a simple randomized algorithm and establish a new bound that depends explicitly on the sampling probabilities. The sampling dependent error bound (i) allows us to better understand the ...
[]
null
15
1505.00526
title_snapshot
[ -0.03685716539621353, -0.01242167130112648, 0.0007441392517648637, 0.030075272545218468, 0.06664013117551804, 0.023129452019929886, 0.03448821231722832, -0.053082846105098724, -0.020475445315241814, -0.06256258487701416, 0.002119754208251834, -0.013136900961399078, -0.06862415373325348, -0...
A Stochastic PCA and SVD Algorithm with an Exponential Convergence Rate
https://proceedings.mlr.press/v37/shamir15.html
[ "Ohad Shamir" ]
null
null
We describe and analyze a simple algorithm for principal component analysis and singular value decomposition, VR-PCA, which uses computationally cheap stochastic iterations, yet converges exponentially fast to the optimal solution. In contrast, existing algorithms suffer either from slow convergence, or computationally...
[]
null
16
1409.2848
title_snapshot
[ 0.004085192456841469, -0.002799991751089692, 0.04601757228374481, 0.04445303976535797, 0.016931500285863876, 0.05403238534927368, 0.03835421800613403, 0.01571843959391117, -0.024419210851192474, -0.06467802077531815, -0.011808004230260849, -0.02534225583076477, -0.060851436108350754, 0.004...
Attribute Efficient Linear Regression with Distribution-Dependent Sampling
https://proceedings.mlr.press/v37/kukliansky15.html
[ "Doron Kukliansky", "Ohad Shamir" ]
null
null
We consider a budgeted learning setting, where the learner can only choose and observe a small subset of the attributes of each training example. We develop efficient algorithms for Ridge and Lasso linear regression, which utilize the geometry of the data by a novel distribution-dependent sampling scheme, and have exce...
[]
null
17
1410.6382
title_judge
[ 0.0012392704375088215, -0.0012028423370793462, 0.0006832602666690946, 0.02500772662460804, 0.060476940125226974, 0.042171698063611984, 0.030037472024559975, -0.03569605201482773, -0.010091920383274555, -0.03916308283805847, -0.0034451764076948166, 0.015160707756876945, -0.0821179449558258, ...
Learning Local Invariant Mahalanobis Distances
https://proceedings.mlr.press/v37/fetaya15.html
[ "Ethan Fetaya", "Shimon Ullman" ]
null
null
For many tasks and data types, there are natural transformations to which the data should be invariant or insensitive. For instance, in visual recognition, natural images should be insensitive to rotation and translation. This requirement and its implications have been important in many machine learning applications, a...
[]
null
18
1502.01176
title_snapshot
[ -0.01284992229193449, -0.0005884008714929223, -0.0031902978662401438, 0.027817700058221817, 0.03840619698166847, 0.05541788786649704, 0.03480792045593262, -0.012453634291887283, -0.015292186290025711, -0.039829425513744354, -0.053313370794057846, -0.0187740046530962, -0.07250124216079712, ...
Finding Linear Structure in Large Datasets with Scalable Canonical Correlation Analysis
https://proceedings.mlr.press/v37/maa15.html
[ "Zhuang Ma", "Yichao Lu", "Dean Foster" ]
null
null
Canonical Correlation Analysis (CCA) is a widely used spectral technique for finding correlation structures in multi-view datasets. In this paper, we tackle the problem of large scale CCA, where classical algorithms, usually requiring computing the product of two huge matrices and huge matrix decomposition, are computa...
[]
null
19
1506.08170
title_snapshot
[ -0.0012855390086770058, -0.0197081808000803, 0.010923512279987335, 0.01693635620176792, 0.044804591685533524, 0.04026612266898155, 0.028546493500471115, 0.014418012462556362, -0.011933010071516037, -0.0399370901286602, -0.013849668204784393, -0.019335534423589706, -0.07827404886484146, 0.0...
Abstraction Selection in Model-based Reinforcement Learning
https://proceedings.mlr.press/v37/jiang15.html
[ "Nan Jiang", "Alex Kulesza", "Satinder Singh" ]
null
null
State abstractions are often used to reduce the complexity of model-based reinforcement learning when only limited quantities of data are available. However, choosing the appropriate level of abstraction is an important problem in practice. Existing approaches have theoretical guarantees only under strong assumptions o...
[]
null
20
null
null
[ -0.06255476176738739, -0.029062163084745407, -0.01060571987181902, 0.05481501296162605, 0.05041331797838211, -0.008014583960175514, 0.00958115141838789, -0.024943286553025246, -0.03587617725133896, -0.0037938908208161592, -0.015868449583649635, 0.024836812168359756, -0.0638444572687149, -0...
Surrogate Functions for Maximizing Precision at the Top
https://proceedings.mlr.press/v37/kar15.html
[ "Purushottam Kar", "Harikrishna Narasimhan", "Prateek Jain" ]
null
null
The problem of maximizing precision at the top of a ranked list, often dubbed Precision@k (prec@k), finds relevance in myriad learning applications such as ranking, multi-label classification, and learning with severe label imbalance. However, despite its popularity, there exist significant gaps in our understanding of...
[]
null
21
1505.06813
title_snapshot
[ -0.023412024602293968, -0.008999754674732685, 0.01221207994967699, 0.05472148209810257, 0.03999100998044014, 0.04083869233727455, 0.01900738663971424, -0.03534587472677231, -0.009667283855378628, -0.026875851675868034, -0.016192490234971046, -0.01966754160821438, -0.05417606979608536, -0.0...
Optimizing Non-decomposable Performance Measures: A Tale of Two Classes
https://proceedings.mlr.press/v37/narasimhana15.html
[ "Harikrishna Narasimhan", "Purushottam Kar", "Prateek Jain" ]
null
null
Modern classification problems frequently present mild to severe label imbalance as well as specific requirements on classification characteristics, and require optimizing performance measures that are non-decomposable over the dataset, such as F-measure. Such measures have spurred much interest and pose specific chall...
[]
null
22
1505.06812
title_snapshot
[ -0.0021003030706197023, -0.009890641085803509, -0.002687707543373108, 0.03281138092279434, 0.04316210746765137, 0.05118648335337639, 0.01500500738620758, -0.02233005501329899, -0.021763844415545464, -0.04217347875237465, -0.008936353027820587, -0.008434507064521313, -0.06354410946369171, -...
Coresets for Nonparametric Estimation - the Case of DP-Means
https://proceedings.mlr.press/v37/bachem15.html
[ "Olivier Bachem", "Mario Lucic", "Andreas Krause" ]
null
null
Scalable training of Bayesian nonparametric models is a notoriously difficult challenge. We explore the use of coresets - a data summarization technique originating from computational geometry - for this task. Coresets are weighted subsets of the data such that models trained on these coresets are provably competitive ...
[]
null
23
null
null
[ -0.02899095043540001, -0.031610891222953796, -0.01219421997666359, 0.050681423395872116, 0.03915736824274063, 0.07277626544237137, 0.01028845738619566, -0.016942333430051804, 0.003064855234697461, -0.06261323392391205, -0.01693326234817505, -0.02371443621814251, -0.06925920397043228, -0.02...
A Relative Exponential Weighing Algorithm for Adversarial Utility-based Dueling Bandits
https://proceedings.mlr.press/v37/gajane15.html
[ "Pratik Gajane", "Tanguy Urvoy", "Fabrice Clérot" ]
null
null
We study the K-armed dueling bandit problem which is a variation of the classical Multi-Armed Bandit (MAB) problem in which the learner receives only relative feedback about the selected pairs of arms. We propose a new algorithm called Relative Exponential-weight algorithm for Exploration and Exploitation (REX3) to han...
[]
null
24
1601.03855
title_snapshot
[ -0.026243634521961212, -0.010360626503825188, 0.016466911882162094, 0.04994535818696022, 0.022364234551787376, 0.004108924884349108, 0.016683658584952354, -0.008333672769367695, -0.02864914946258068, -0.03844821825623512, -0.021377768367528915, 0.030224615707993507, -0.05901121348142624, -...
Functional Subspace Clustering with Application to Time Series
https://proceedings.mlr.press/v37/bahadori15.html
[ "Mohammad Taha Bahadori", "David Kale", "Yingying Fan", "Yan Liu" ]
null
null
Functional data, where samples are random functions, are increasingly common and important in a variety of applications, such as health care and traffic analysis. They are naturally high dimensional and lie along complex manifolds. These properties warrant use of the subspace assumption, but most state-of-the-art subsp...
[]
null
25
null
null
[ -0.008337758481502533, -0.04137115553021431, 0.030879750847816467, 0.019301077350974083, 0.058439627289772034, 0.054282039403915405, 0.034590162336826324, -0.004985218867659569, 0.008590077050030231, -0.05284988880157471, 0.00868478324264288, -0.005316548515111208, -0.06322300434112549, 0....
Accelerated Online Low Rank Tensor Learning for Multivariate Spatiotemporal Streams
https://proceedings.mlr.press/v37/yua15.html
[ "Rose Yu", "Dehua Cheng", "Yan Liu" ]
null
null
Low-rank tensor learning has many applications in machine learning. A series of batch learning algorithms have achieved great successes. However, in many emerging applications, such as climate data analysis, we are confronted with large-scale tensor streams, which poses significant challenges to existing solution in te...
[]
null
26
null
null
[ -0.03752949833869934, -0.055842332541942596, 0.04297488555312157, 0.027881182730197906, 0.009024207480251789, 0.03311218321323395, 0.01320907287299633, 0.0023315446451306343, -0.03929368406534195, -0.04833947867155075, -0.021647265180945396, 0.003977643791586161, -0.06406429409980774, 0.02...
Atomic Spatial Processes
https://proceedings.mlr.press/v37/jewell15.html
[ "Sean Jewell", "Neil Spencer", "Alexandre Bouchard-Côté" ]
null
null
The emergence of compact GPS systems and the establishment of open data initiatives has resulted in widespread availability of spatial data for many urban centres. These data can be leveraged to develop data-driven intelligent resource allocation systems for urban issues such as policing, sanitation, and transportation...
[]
null
27
null
null
[ 0.02963920868933201, 0.014029793441295624, -0.0175272636115551, 0.03154753893613815, 0.0326024666428566, 0.04614352807402611, 0.00801685731858015, 0.031186366453766823, 0.0037256621289998293, -0.062322720885276794, -0.02335280552506447, -0.038992300629615784, -0.07681666314601898, 0.001595...
Classification with Low Rank and Missing Data
https://proceedings.mlr.press/v37/hazan15.html
[ "Elad Hazan", "Roi Livni", "Yishay Mansour" ]
null
null
We consider classification and regression tasks where we have missing data and assume that the (clean) data resides in a low rank subspace. Finding a hidden subspace is known to be computationally hard. Nevertheless, using a non-proper formulation we give an efficient agnostic algorithm that classifies as good as the b...
[]
null
28
1501.03273
title_snapshot
[ -0.03378327563405037, -0.0185895673930645, 0.020994018763303757, 0.05494452267885208, 0.05415775999426842, 0.011802279390394688, 0.0178199615329504, -0.04303567111492157, -0.045462869107723236, -0.02229108102619648, -0.03908487409353256, 0.029240768402814865, -0.06001967191696167, 0.018145...
Dynamic Sensing: Better Classification under Acquisition Constraints
https://proceedings.mlr.press/v37/richman15.html
[ "Oran Richman", "Shie Mannor" ]
null
null
In many machine learning applications the quality of the data is limited by resource constraints (may it be power, bandwidth, memory, ...). In such cases, the constraints are on the average resources allocated, therefore there is some control over each sample’s quality. In most cases this option remains unused and the ...
[]
null
29
null
null
[ -0.02227305807173252, -0.023570677265524864, -0.011063543148338795, 0.032361775636672974, 0.05185369402170181, 0.01962762512266636, 0.01642942801117897, -0.012283381074666977, -0.047729671001434326, -0.054894186556339264, -0.015984144061803818, 0.0037234900519251823, -0.09371934086084366, ...
A Modified Orthant-Wise Limited Memory Quasi-Newton Method with Convergence Analysis
https://proceedings.mlr.press/v37/gonga15.html
[ "Pinghua Gong", "Jieping Ye" ]
null
null
The Orthant-Wise Limited memory Quasi-Newton (OWL-QN) method has been demonstrated to be very effective in solving the \ell_1-regularized sparse learning problem. OWL-QN extends the L-BFGS from solving unconstrained smooth optimization problems to \ell_1-regularized (non-smooth) sparse learning problems. At each iterat...
[]
null
30
null
null
[ -0.05850416421890259, -0.020140618085861206, 0.030331464484333992, 0.020189126953482628, 0.0539417564868927, 0.03173470124602318, -0.009811460971832275, 0.027139276266098022, -0.052051592618227005, -0.02593611180782318, -0.003713271114975214, 0.009278797544538975, -0.07803080976009369, -0....
Telling cause from effect in deterministic linear dynamical systems
https://proceedings.mlr.press/v37/shajarisales15.html
[ "Naji Shajarisales", "Dominik Janzing", "Bernhard Schoelkopf", "Michel Besserve" ]
null
null
Telling a cause from its effect using observed time series data is a major challenge in natural and social sciences. Assuming the effect is generated by the cause through a linear system, we propose a new approach based on the hypothesis that nature chooses the “cause” and the “mechanism generating the effect from the ...
[]
null
31
1503.01299
title_snapshot
[ -0.00004355722921900451, -0.021389007568359375, -0.021349968388676643, 0.01058865524828434, 0.04429193213582039, 0.051780324429273605, 0.04751898720860481, 0.02534535527229309, -0.03745495527982712, -0.06536535918712616, 0.01815461926162243, 0.00461255619302392, -0.054365091025829315, 0.00...
High Dimensional Bayesian Optimisation and Bandits via Additive Models
https://proceedings.mlr.press/v37/kandasamy15.html
[ "Kirthevasan Kandasamy", "Jeff Schneider", "Barnabas Poczos" ]
null
null
Bayesian Optimisation (BO) is a technique used in optimising a D-dimensional function which is typically expensive to evaluate. While there have been many successes for BO in low dimensions, scaling it to high dimensions has been notoriously difficult. Existing literature on the topic are under very restrictive setting...
[]
null
32
1503.01673
title_snapshot
[ -0.018536902964115143, 0.01824570633471012, 0.018147002905607224, 0.0136609748005867, 0.004306880757212639, 0.06917241960763931, 0.02730848267674446, -0.027208251878619194, -0.021750211715698242, -0.0404072031378746, -0.019126370549201965, 0.0027112814132124186, -0.06937333196401596, -0.00...
Theory of Dual-sparse Regularized Randomized Reduction
https://proceedings.mlr.press/v37/yangb15.html
[ "Tianbao Yang", "Lijun Zhang", "Rong Jin", "Shenghuo Zhu" ]
null
null
In this paper, we study randomized reduction methods, which reduce high-dimensional features into low-dimensional space by randomized methods (e.g., random projection, random hashing), for large-scale high-dimensional classification. Previous theoretical results on randomized reduction methods hinge on strong assumptio...
[]
null
33
1504.03991
title_snapshot
[ -0.009935096837580204, -0.024469735100865364, -0.007086955942213535, 0.03619153052568436, 0.031307823956012726, 0.046462222933769226, -0.0027115505654364824, -0.011194099672138691, -0.027312472462654114, -0.06116452068090439, -0.0039452966302633286, -0.01964355632662773, -0.05203054472804069...
Generalization error bounds for learning to rank: Does the length of document lists matter?
https://proceedings.mlr.press/v37/tewari15.html
[ "Ambuj Tewari", "Sougata Chaudhuri" ]
null
null
We consider the generalization ability of algorithms for learning to rank at a query level, a problem also called subset ranking. Existing generalization error bounds necessarily degrade as the size of the document list associated with a query increases. We show that such a degradation is not intrinsic to the problem. ...
[]
null
34
1603.01860
title_snapshot
[ -0.01342430803924799, -0.02703472413122654, -0.007970324717462063, 0.04229915514588356, 0.025579821318387985, -0.014355897903442383, 0.03272346407175064, -0.03107527829706669, -0.045909982174634933, -0.005658247042447329, -0.023611828684806824, 0.02611488103866577, -0.0784643366932869, -0....
PeakSeg: constrained optimal segmentation and supervised penalty learning for peak detection in count data
https://proceedings.mlr.press/v37/hocking15.html
[ "Toby Hocking", "Guillem Rigaill", "Guillaume Bourque" ]
null
null
Peak detection is a central problem in genomic data analysis, and current algorithms for this task are unsupervised and mostly effective for a single data type and pattern (e.g. H3K4me3 data with a sharp peak pattern). We propose PeakSeg, a new constrained maximum likelihood segmentation model for peak detection with a...
[]
null
35
null
null
[ -0.03813531622290611, -0.023973051458597183, -0.04597765579819679, 0.010969417169690132, 0.02570364437997341, -0.009362825192511082, 0.03675417602062225, -0.014845783822238445, -0.03785230591893196, -0.02372702583670616, -0.01263831090182066, 0.0027722700033336878, -0.06415244191884995, 0....
Mind the duality gap: safer rules for the Lasso
https://proceedings.mlr.press/v37/fercoq15.html
[ "Olivier Fercoq", "Alexandre Gramfort", "Joseph Salmon" ]
null
null
Screening rules allow to early discard irrelevant variables from the optimization in Lasso problems, or its derivatives, making solvers faster. In this paper, we propose new versions of the so-called \textitsafe rules for the Lasso. Based on duality gap considerations, our new rules create safe test regions whose diame...
[]
null
36
1505.03410
title_snapshot
[ -0.03496488556265831, -0.013918391428887844, 0.01749161258339882, 0.015851614996790886, 0.06198998913168907, 0.016412248834967613, 0.043656375259160995, -0.023056620731949806, -0.04317087307572365, -0.04807765781879425, -0.0016351148951798677, 0.015565590932965279, -0.07066212594509125, 0....
A General Analysis of the Convergence of ADMM
https://proceedings.mlr.press/v37/nishihara15.html
[ "Robert Nishihara", "Laurent Lessard", "Ben Recht", "Andrew Packard", "Michael Jordan" ]
null
null
We provide a new proof of the linear convergence of the alternating direction method of multipliers (ADMM) when one of the objective terms is strongly convex. Our proof is based on a framework for analyzing optimization algorithms introduced in Lessard et al. (2014), reducing algorithm convergence to verifying the stab...
[]
null
37
1502.02009
title_snapshot
[ -0.05107327550649643, -0.015402412973344326, 0.019926000386476517, 0.013394162058830261, 0.015386064536869526, 0.04031268507242203, 0.01943458616733551, 0.0014132439391687512, -0.049768976867198944, -0.03101142682135105, -0.011879515834152699, -0.012003835290670395, -0.057895008474588394, ...
Stochastic Primal-Dual Coordinate Method for Regularized Empirical Risk Minimization
https://proceedings.mlr.press/v37/zhanga15.html
[ "Yuchen Zhang", "Xiao Lin" ]
null
null
We consider a generic convex optimization problem associated with regularized empirical risk minimization of linear predictors. The problem structure allows us to reformulate it as a convex-concave saddle point problem. We propose a stochastic primal-dual coordinate method, which alternates between maximizing over one ...
[]
null
38
1409.3257
title_snapshot
[ -0.032556332647800446, -0.017000431194901466, -0.023001842200756073, 0.04539540782570839, 0.03527709096670151, 0.06655929237604141, -0.0034706576261669397, -0.01022193394601345, -0.012197514995932579, -0.05486152693629265, -0.006168706342577934, -0.004982493352144957, -0.0386282242834568, ...
DiSCO: Distributed Optimization for Self-Concordant Empirical Loss
https://proceedings.mlr.press/v37/zhangb15.html
[ "Yuchen Zhang", "Xiao Lin" ]
null
null
We propose a new distributed algorithm for empirical risk minimization in machine learning. The algorithm is based on an inexact damped Newton method, where the inexact Newton steps are computed by a distributed preconditioned conjugate gradient method. We analyze its iteration complexity and communication efficiency f...
[]
null
39
1501.00263
title_judge
[ -0.04130466654896736, -0.030670063570141792, 0.0019162703538313508, 0.05757419392466545, 0.04547698423266411, 0.04181162267923355, 0.021944450214505196, -0.029067520052194595, -0.02975766733288765, -0.034254953265190125, -0.007815725170075893, 0.027681035920977592, -0.055373162031173706, -...
Spectral MLE: Top-K Rank Aggregation from Pairwise Comparisons
https://proceedings.mlr.press/v37/chena15.html
[ "Yuxin Chen", "Changho Suh" ]
null
null
This paper explores the preference-based top-K rank aggregation problem. Suppose that a collection of items is repeatedly compared in pairs, and one wishes to recover a consistent ordering that emphasizes the top-K ranked items, based on partially revealed preferences. We focus on the Bradley-Terry-Luce (BTL) model tha...
[]
null
40
1504.07218
title_snapshot
[ -0.01614784449338913, -0.004778034053742886, 0.038158614188432693, 0.026884427294135094, 0.019421570003032684, -0.015602771192789078, 0.025327323004603386, -0.006132916547358036, -0.03484323248267174, -0.030747272074222565, -0.04484650120139122, 0.0023560018744319677, -0.06223803758621216, ...
Paired-Dual Learning for Fast Training of Latent Variable Hinge-Loss MRFs
https://proceedings.mlr.press/v37/bach15.html
[ "Stephen Bach", "Bert Huang", "Jordan Boyd-Graber", "Lise Getoor" ]
null
null
Latent variables allow probabilistic graphical models to capture nuance and structure in important domains such as network science, natural language processing, and computer vision. Naive approaches to learning such complex models can be prohibitively expensive—because they require repeated inferences to update beliefs...
[]
null
41
null
null
[ 0.014370420947670937, 0.00880456529557705, -0.01850755698978901, 0.05238780006766319, 0.013938786461949348, 0.034748468548059464, 0.024036811664700508, -0.008371084928512573, -0.019312642514705658, -0.03968396037817001, 0.006867315620183945, 0.02339046448469162, -0.05606906861066818, -0.01...
Structural Maxent Models
https://proceedings.mlr.press/v37/cortes15.html
[ "Corinna Cortes", "Vitaly Kuznetsov", "Mehryar Mohri", "Umar Syed" ]
null
null
We present a new class of density estimation models, Structural Maxent models, with feature functions selected from possibly very complex families. The design of our models is motivated by data-dependent convergence bounds and benefits from new data-dependent learning bounds expressed in terms of the Rademacher complex...
[]
null
42
null
null
[ -0.01601484790444374, -0.011001816019415855, 0.01936234161257744, 0.06302714347839355, 0.033931225538253784, 0.04536951705813408, 0.002202654257416725, -0.016203507781028748, -0.017162276431918144, -0.040136758238077164, -0.010817719623446465, 0.004774930886924267, -0.06186316907405853, -0...
A Provable Generalized Tensor Spectral Method for Uniform Hypergraph Partitioning
https://proceedings.mlr.press/v37/ghoshdastidar15.html
[ "Debarghya Ghoshdastidar", "Ambedkar Dukkipati" ]
null
null
Matrix spectral methods play an important role in statistics and machine learning, and most often the word ‘matrix’ is dropped as, by default, one assumes that similarities or affinities are measured between two points, thereby resulting in similarity matrices. However, recent challenges in computer vision and text min...
[]
null
43
null
null
[ 0.00001784438245522324, -0.02281474880874157, 0.013356484472751617, 0.021166665479540825, 0.015457459725439548, 0.001532075461000204, 0.02462918870151043, -0.0341000109910965, -0.0063229636289179325, -0.08057180792093277, -0.006740150041878223, 0.010164842009544373, -0.08290757238864899, 0...
The Benefits of Learning with Strongly Convex Approximate Inference
https://proceedings.mlr.press/v37/london15.html
[ "Ben London", "Bert Huang", "Lise Getoor" ]
null
null
We explore the benefits of strongly convex free energies in variational inference, providing both theoretical motivation and a new meta-algorithm. Using the duality between strong convexity and stability, we prove a high-probability bound on the error of learned marginals that is inversely proportional to the modulus o...
[]
null
44
null
null
[ -0.02692054584622383, 0.0026964384596794844, 0.007327111903578043, 0.05277859419584274, 0.03667953610420227, 0.01257187407463789, 0.016653086990118027, 0.02601483464241028, -0.027806205675005913, -0.033217333257198334, -0.0073203155770897865, 0.030531879514455795, -0.07295303046703339, 0.0...
Pushing the Limits of Affine Rank Minimization by Adapting Probabilistic PCA
https://proceedings.mlr.press/v37/xin15.html
[ "Bo Xin", "David Wipf" ]
null
null
Many applications require recovering a matrix of minimal rank within an affine constraint set, with matrix completion a notable special case. Because the problem is NP-hard in general, it is common to replace the matrix rank with the nuclear norm, which acts as a convenient convex surrogate. While elegant theoretical c...
[]
null
45
1406.2504
title_judge
[ -0.009122702293097973, -0.027268612757325172, 0.017369724810123444, 0.03724737465381622, 0.025741824880242348, 0.011595441028475761, 0.020280756056308746, -0.009174593724310398, -0.05162118375301361, -0.04123581200838089, -0.043435633182525635, -0.005679463502019644, -0.061094556003808975, ...
Budget Allocation Problem with Multiple Advertisers: A Game Theoretic View
https://proceedings.mlr.press/v37/maehara15.html
[ "Takanori Maehara", "Akihiro Yabe", "Ken-ichi Kawarabayashi" ]
null
null
In marketing planning, advertisers seek to maximize the number of customers by allocating given budgets to each media channel effectively. The budget allocation problem with a bipartite influence model captures this scenario; however, the model is problematic because it assumes there is only one advertiser in the marke...
[]
null
46
null
null
[ -0.022996950894594193, 0.005714071914553642, -0.012636224739253521, 0.004678964149206877, 0.02141606993973255, 0.03931594640016556, 0.003458963939920068, 0.018314259126782417, -0.02209208905696869, -0.038178153336048126, -0.0040452913381159306, 0.016137612983584404, -0.05442268028855324, -...
Tracking Approximate Solutions of Parameterized Optimization Problems over Multi-Dimensional (Hyper-)Parameter Domains
https://proceedings.mlr.press/v37/blechschmidt15.html
[ "Katharina Blechschmidt", "Joachim Giesen", "Soeren Laue" ]
null
null
Many machine learning methods are given as parameterized optimization problems. Important examples of such parameters are regularization- and kernel hyperparameters. These parameters have to be tuned carefully since the choice of their values can have a significant impact on the statistical performance of the learning ...
[]
null
47
null
null
[ -0.05025052651762962, -0.012605754658579826, 0.022748146206140518, 0.032741669565439224, 0.05540192872285843, 0.043656691908836365, 0.024581510573625565, -0.020403748378157616, -0.024195265024900436, -0.051873207092285156, -0.02116594836115837, 0.0017666197381913662, -0.04109217971563339, ...
Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
https://proceedings.mlr.press/v37/ioffe15.html
[ "Sergey Ioffe", "Christian Szegedy" ]
null
null
Training Deep Neural Networks is complicated by the fact that the distribution of each layer’s inputs changes during training, as the parameters of the previous layers change. This slows down the training by requiring lower learning rates and careful parameter initialization, and makes it notoriously hard to train mode...
[]
null
48
1502.03167
title_snapshot
[ -0.023044951260089874, -0.04672502726316452, -0.01715160347521305, 0.03309815749526024, 0.040800489485263824, 0.05611713230609894, 0.028435608372092247, -0.0014761603670194745, -0.001898545422591269, -0.046081602573394775, -0.020928869023919106, -0.007995790801942348, -0.046336326748132706, ...
Distributed Estimation of Generalized Matrix Rank: Efficient Algorithms and Lower Bounds
https://proceedings.mlr.press/v37/zhangc15.html
[ "Yuchen Zhang", "Martin Wainwright", "Michael Jordan" ]
null
null
We study the following generalized matrix rank estimation problem: given an n-by-n matrix and a constant c > 0, estimate the number of eigenvalues that are greater than c. In the distributed setting, the matrix of interest is the sum of m matrices held by separate machines. We show that any deterministic algorithm solv...
[]
null
49
1502.01403
title_snapshot
[ -0.036164604127407074, -0.01794314943253994, 0.02006264217197895, 0.03324133902788162, 0.01786804012954235, 0.01330794207751751, 0.034278515726327896, 0.0030565066263079643, -0.02005479857325554, -0.048422981053590775, 0.006923515349626541, -0.0099715581163764, -0.07883261144161224, -0.011...
Landmarking Manifolds with Gaussian Processes
https://proceedings.mlr.press/v37/liang15.html
[ "Dawen Liang", "John Paisley" ]
null
null
We present an algorithm for finding landmarks along a manifold. These landmarks provide a small set of locations spaced out along the manifold such that they capture the low-dimensional non-linear structure of the data embedded in the high-dimensional space. The approach does not select points directly from the dataset...
[]
null
50
null
null
[ -0.011636718176305294, -0.006962195970118046, 0.0018493304960429668, 0.020978637039661407, -0.0029405716340988874, 0.03471246361732483, 0.0339277945458889, 0.006670054979622364, -0.033151738345623016, -0.04133165627717972, -0.019910918548703194, -0.0028266292065382004, -0.08840154111385345, ...
Markov Mixed Membership Models
https://proceedings.mlr.press/v37/zhangd15.html
[ "Aonan Zhang", "John Paisley" ]
null
null
We present a Markov mixed membership model (Markov M3) for grouped data that learns a fully connected graph structure among mixing components. A key feature of Markov M3 is that it interprets the mixed membership assignment as a Markov random walk over this graph of nodes. This is in contrast to tree-structured models ...
[]
null
51
null
null
[ 0.011423581279814243, -0.016651134938001633, 0.001364603522233665, 0.04936479032039642, 0.03443215414881706, 0.04892294108867645, 0.0401046946644783, -0.0005091733764857054, -0.03238288313150406, -0.010574555024504662, 0.001328483340330422, -0.0009122712071985006, -0.07205817848443985, -0....
A Unified Framework for Outlier-Robust PCA-like Algorithms
https://proceedings.mlr.press/v37/yangc15.html
[ "Wenzhuo Yang", "Huan Xu" ]
null
null
We propose a unified framework for making a wide range of PCA-like algorithms – including the standard PCA, sparse PCA and non-negative sparse PCA, etc. – robust when facing a constant fraction of arbitrarily corrupted outliers. Our theoretic analysis establishes solid performance guarantees of the proposed framework: ...
[]
null
52
null
null
[ 0.016510486602783203, -0.03700253739953041, 0.015783807262778282, 0.035353127866983414, 0.051256097853183746, 0.02532092109322548, 0.010193753987550735, -0.001261250232346356, -0.053903594613075256, -0.0474817156791687, -0.033539067953825, -0.034606970846652985, -0.08071541786193848, -0.01...
Streaming Sparse Principal Component Analysis
https://proceedings.mlr.press/v37/yangd15.html
[ "Wenzhuo Yang", "Huan Xu" ]
null
null
This paper considers estimating the leading k principal components with at most s non-zero attributes from p-dimensional samples collected sequentially in memory limited environments. We develop and analyze two memory and computational efficient algorithms called streaming sparse PCA and streaming sparse ECA for analyz...
[]
null
53
null
null
[ -0.02738066203892231, -0.0322890467941761, 0.019245509058237076, 0.037445783615112305, 0.018768085166811943, 0.040078744292259216, 0.013419640250504017, 0.025254279375076294, -0.04387526959180832, -0.03467116504907608, -0.01088507566601038, -0.03304239735007286, -0.08068080246448517, -0.00...
A Divide and Conquer Framework for Distributed Graph Clustering
https://proceedings.mlr.press/v37/yange15.html
[ "Wenzhuo Yang", "Huan Xu" ]
null
null
Graph clustering is about identifying clusters of closely connected nodes, and is a fundamental technique of data analysis with many applications including community detection, VLSI network partitioning, collaborative filtering, and many others. In order to improve the scalability of existing graph clustering algorithm...
[]
null
54
null
null
[ 0.0005568733322434127, -0.01431175135076046, 0.01231249887496233, 0.04079626873135567, 0.05176598206162453, 0.03521263599395752, 0.011806945316493511, -0.011622879654169083, -0.026554547250270844, -0.04028446972370148, 0.021690336987376213, -0.04148409888148308, -0.07290284335613251, 0.020...
How Can Deep Rectifier Networks Achieve Linear Separability and Preserve Distances?
https://proceedings.mlr.press/v37/an15.html
[ "Senjian An", "Farid Boussaid", "Mohammed Bennamoun" ]
null
null
This paper investigates how hidden layers of deep rectifier networks are capable of transforming two or more pattern sets to be linearly separable while preserving the distances with a guaranteed degree, and proves the universal classification power of such distance preserving rectifier networks. Through the nearly iso...
[]
null
55
null
null
[ -0.004602292086929083, -0.008463867008686066, 0.002078753663226962, 0.0013732848456129432, 0.06437212973833084, 0.01027870923280716, 0.018130797892808914, -0.033615756779909134, -0.0294648390263319, -0.03185582906007767, 0.041612181812524796, -0.009586594998836517, -0.06689827889204025, 0....
Improved Regret Bounds for Undiscounted Continuous Reinforcement Learning
https://proceedings.mlr.press/v37/lakshmanan15.html
[ "K. Lakshmanan", "Ronald Ortner", "Daniil Ryabko" ]
null
null
We consider the problem of undiscounted reinforcement learning in continuous state space. Regret bounds in this setting usually hold under various assumptions on the structure of the reward and transition function. Under the assumption that the rewards and transition probabilities are Lipschitz, for 1-dimensional state...
[]
null
56
null
null
[ -0.057523008435964584, -0.02869214117527008, -0.00801530946046114, 0.050288811326026917, 0.05413687229156494, 0.028798457235097885, 0.02422860637307167, 0.01359387207776308, -0.007603928912431002, -0.041581809520721436, -0.010688385926187038, 0.022625837475061417, -0.06452348828315735, -0....
The Fundamental Incompatibility of Scalable Hamiltonian Monte Carlo and Naive Data Subsampling
https://proceedings.mlr.press/v37/betancourt15.html
[ "Michael Betancourt" ]
null
null
Leveraging the coherent exploration of Hamiltonian flow, Hamiltonian Monte Carlo produces computationally efficient Monte Carlo estimators, even with respect to complex and high-dimensional target distributions. When confronted with data-intensive applications, however, the algorithm may be too expensive to implement, ...
[]
null
57
1502.01510
title_judge
[ -0.022392023354768753, 0.005724998190999031, -0.017777232453227043, 0.08061283826828003, 0.037427715957164764, 0.002482791431248188, 0.030432168394327164, -0.030399125069379807, -0.018518131226301193, -0.066610187292099, 0.024331580847501755, -0.026602506637573242, -0.07047480344772339, 0....
Faster Rates for the Frank-Wolfe Method over Strongly-Convex Sets
https://proceedings.mlr.press/v37/garbera15.html
[ "Dan Garber", "Elad Hazan" ]
null
null
The Frank-Wolfe method (a.k.a. conditional gradient algorithm) for smooth optimization has regained much interest in recent years in the context of large scale optimization and machine learning. A key advantage of the method is that it avoids projections - the computational bottleneck in many applications - replacing i...
[]
null
58
1406.1305
title_snapshot
[ -0.030254963785409927, -0.011254309676587582, 0.03875641152262688, 0.0213441401720047, 0.027622485533356667, 0.03369396924972534, 0.018881885334849358, 0.005768140312284231, 0.00160258321557194, -0.04771159961819649, -0.008527536876499653, 0.013323679566383362, -0.06046703830361366, -0.010...
Ordered Stick-Breaking Prior for Sequential MCMC Inference of Bayesian Nonparametric Models
https://proceedings.mlr.press/v37/das15.html
[ "Mrinal Das", "Trapit Bansal", "Chiranjib Bhattacharyya" ]
null
null
This paper introduces ordered stick-breaking process (OSBP), where the atoms in a stick-breaking process (SBP) appear in order. The choice of weights on the atoms of OSBP ensure that; (1) probability of adding new atoms exponentially decrease, and (2) OSBP, though non-exchangeable, admit predictive probability function...
[]
null
59
null
null
[ -0.03789927065372467, -0.007053588982671499, -0.03566670045256615, 0.028567887842655182, 0.009501495398581028, 0.0409068688750267, 0.014969931915402412, 0.013620763085782528, -0.021653879433870316, -0.05689089000225067, 0.013328565284609795, 0.017033370211720467, -0.05472218990325928, -0.0...
Online Learning of Eigenvectors
https://proceedings.mlr.press/v37/garberb15.html
[ "Dan Garber", "Elad Hazan", "Tengyu Ma" ]
null
null
Computing the leading eigenvector of a symmetric real matrix is a fundamental primitive of numerical linear algebra with numerous applications. We consider a natural online extension of the leading eigenvector problem: a sequence of matrices is presented and the goal is to predict for each matrix a unit vector, with th...
[]
null
60
null
null
[ -0.0402451753616333, -0.01634831354022026, 0.03652355447411537, 0.022321313619613647, 0.02185484953224659, 0.03355616703629494, 0.024886613711714745, 0.020791536197066307, -0.007058264687657356, -0.04781915992498398, -0.008034349419176579, -0.008200274780392647, -0.0700804591178894, -0.011...
A Unifying Framework of Anytime Sparse Gaussian Process Regression Models with Stochastic Variational Inference for Big Data
https://proceedings.mlr.press/v37/hoang15.html
[ "Trong Nghia Hoang", "Quang Minh Hoang", "Bryan Kian Hsiang Low" ]
null
null
This paper presents a novel unifying framework of anytime sparse Gaussian process regression (SGPR) models that can produce good predictive performance fast and improve their predictive performance over time. Our proposed unifying framework reverses the variational inference procedure to theoretically construct a non-t...
[]
null
61
null
null
[ -0.009400059469044209, -0.01976884715259075, 0.011596235446631908, 0.01840665005147457, 0.039271630346775055, 0.04573037847876549, 0.029085341840982437, 0.023495392873883247, -0.03317844122648239, -0.038729917258024216, -0.011770068667829037, -0.006332265213131905, -0.06660540401935577, 0....
Yinyang K-Means: A Drop-In Replacement of the Classic K-Means with Consistent Speedup
https://proceedings.mlr.press/v37/ding15.html
[ "Yufei Ding", "Yue Zhao", "Xipeng Shen", "Madanlal Musuvathi", "Todd Mytkowicz" ]
null
null
This paper presents Yinyang K-means, a new algorithm for K-means clustering. By clustering the centers in the initial stage, and leveraging efficiently maintained lower and upper bounds between a point and centers, it more effectively avoids unnecessary distance calculations than prior algorithms. It significantly outp...
[]
null
62
null
null
[ -0.00595424510538578, -0.01278231292963028, 0.018871985375881195, 0.015957793220877647, 0.03919446840882301, 0.052486881613731384, 0.004500453360378742, 0.004480084404349327, -0.001974931452423334, -0.03703884035348892, 0.0007360243471339345, -0.04028581455349922, -0.03881657496094704, 0.0...
Ordinal Mixed Membership Models
https://proceedings.mlr.press/v37/virtanen15.html
[ "Seppo Virtanen", "Mark Girolami" ]
null
null
We present a novel class of mixed membership models for joint distributions of groups of observations that co-occur with ordinal response variables for each group for learning statistical associations between the ordinal response variables and the observation groups. The class of proposed models addresses a requirement...
[]
null
63
null
null
[ 0.007411826401948929, -0.0016094858292490244, 0.007310609798878431, 0.012856762856245041, 0.03356599435210228, 0.044086575508117676, 0.04971751570701599, -0.015177397057414055, -0.034385841339826584, 0.004928318317979574, -0.0069377184845507145, 0.009823485277593136, -0.06613989919424057, ...
Online Tracking by Learning Discriminative Saliency Map with Convolutional Neural Network
https://proceedings.mlr.press/v37/hong15.html
[ "Seunghoon Hong", "Tackgeun You", "Suha Kwak", "Bohyung Han" ]
null
null
We propose an online visual tracking algorithm by learning discriminative saliency map using Convolutional Neural Network (CNN). Given a CNN pre-trained on a large-scale image repository in offline, our algorithm takes outputs from hidden layers of the network as feature descriptors since they show excellent representa...
[]
null
64
1502.06796
title_snapshot
[ 0.018672147765755653, -0.01123329158872366, 0.04808523878455162, 0.03268542140722275, 0.02257067710161209, 0.021385423839092255, 0.005626251455396414, 0.04238077998161316, -0.029199182987213135, -0.047509510070085526, -0.04590006545186043, -0.005737638100981712, -0.04393523931503296, -0.02...
Fast Kronecker Inference in Gaussian Processes with non-Gaussian Likelihoods
https://proceedings.mlr.press/v37/flaxman15.html
[ "Seth Flaxman", "Andrew Wilson", "Daniel Neill", "Hannes Nickisch", "Alex Smola" ]
null
null
Gaussian processes (GPs) are a flexible class of methods with state of the art performance on spatial statistics applications. However, GPs require O(n^3) computations and O(n^2) storage, and popular GP kernels are typically limited to smoothing and interpolation. To address these difficulties, Kronecker methods have b...
[]
null
65
null
null
[ 0.0048846774734556675, -0.049060482531785965, 0.01871033012866974, 0.04206471890211105, 0.02701161801815033, 0.04464203119277954, 0.015266504138708115, 0.010418017394840717, -0.023676672950387, -0.05350268632173538, -0.001370481913909316, 0.00028277680394239724, -0.05454714223742485, 0.021...
Statistical and Algorithmic Perspectives on Randomized Sketching for Ordinary Least-Squares
https://proceedings.mlr.press/v37/raskutti15.html
[ "Garvesh Raskutti", "Michael Mahoney" ]
null
null
We consider statistical and algorithmic aspects of solving large-scale least-squares (LS) problems using randomized sketching algorithms. Prior results show that, from an \emphalgorithmic perspective, when using sketching matrices constructed from random projections and leverage-score sampling, if the number of samples...
[]
null
66
1505.06659
title_judge
[ -0.0005552132497541606, -0.02030966244637966, 0.010947825387120247, 0.03995322808623314, 0.05360208451747894, 0.03690328076481819, 0.021703481674194336, 0.0019104554085060954, -0.03857647255063057, -0.06617588549852371, -0.029026588425040245, -0.04150691255927086, -0.07080141454935074, -0....
On TD(0) with function approximation: Concentration bounds and a centered variant with exponential convergence
https://proceedings.mlr.press/v37/korda15.html
[ "Nathaniel Korda", "Prashanth La" ]
null
null
We provide non-asymptotic bounds for the well-known temporal difference learning algorithm TD(0) with linear function approximators. These include high-probability bounds as well as bounds in expectation. Our analysis suggests that a step-size inversely proportional to the number of iterations cannot guarantee optimal ...
[]
null
67
1411.3224
title_snapshot
[ -0.006272341590374708, -0.004822836257517338, -0.031047683209180832, 0.03506476804614067, 0.039987917989492416, 0.01860084943473339, 0.030630720779299736, 0.005310026463121176, -0.017966091632843018, -0.011570069938898087, 0.025265375152230263, -0.007323926780372858, -0.07425682991743088, ...
Learning Parametric-Output HMMs with Two Aliased States
https://proceedings.mlr.press/v37/weiss15.html
[ "Roi Weiss", "Boaz Nadler" ]
null
null
In various applications involving hidden Markov models (HMMs), some of the hidden states are aliased, having identical output distributions. The minimality, identifiability and learnability of such aliased HMMs have been long standing problems, with only partial solutions provided thus far. In this paper we focus on pa...
[]
null
68
1502.02158
title_snapshot
[ -0.019843123853206635, 0.029343830421566963, -0.01347815990447998, -0.0019004150526598096, 0.05264498293399811, 0.07233992218971252, 0.0490497425198555, 0.021219298243522644, -0.015520887449383736, -0.03832966461777687, 0.01092703640460968, 0.014293338172137737, -0.05628940835595131, 0.003...
Latent Gaussian Processes for Distribution Estimation of Multivariate Categorical Data
https://proceedings.mlr.press/v37/gala15.html
[ "Yarin Gal", "Yutian Chen", "Zoubin Ghahramani" ]
null
null
Multivariate categorical data occur in many applications of machine learning. One of the main difficulties with these vectors of categorical variables is sparsity. The number of possible observations grows exponentially with vector length, but dataset diversity might be poor in comparison. Recent models have gained sig...
[]
null
69
1503.02182
title_snapshot
[ 0.013969986699521542, -0.017916342243552208, -0.015493882820010185, 0.042855046689510345, 0.03932637348771095, 0.03669664263725281, 0.02040659822523594, 0.004600603599101305, -0.030023647472262383, -0.03403153270483017, -0.022692983970046043, -0.00455182371661067, -0.05084199085831642, 0.0...
Improving the Gaussian Process Sparse Spectrum Approximation by Representing Uncertainty in Frequency Inputs
https://proceedings.mlr.press/v37/galb15.html
[ "Yarin Gal", "Richard Turner" ]
null
null
Standard sparse pseudo-input approximations to the Gaussian process (GP) cannot handle complex functions well. Sparse spectrum alternatives attempt to answer this but are known to over-fit. We suggest the use of variational inference for the sparse spectrum approximation to avoid both issues. We model the covariance fu...
[]
null
70
1503.02424
title_snapshot
[ 0.0019062402425333858, 0.002702887635678053, 0.014414481818675995, 0.022903388366103172, 0.0442088358104229, 0.03995945304632187, 0.019081907346844673, -0.00829195324331522, -0.02913839742541313, -0.05287446081638336, 0.006389453075826168, 0.03369656205177307, -0.08014577627182007, 0.02858...
Ranking from Stochastic Pairwise Preferences: Recovering Condorcet Winners and Tournament Solution Sets at the Top
https://proceedings.mlr.press/v37/rajkumar15.html
[ "Arun Rajkumar", "Suprovat Ghoshal", "Lek-Heng Lim", "Shivani Agarwal" ]
null
null
We consider the problem of ranking n items from stochastically sampled pairwise preferences. It was shown recently that when the underlying pairwise preferences are acyclic, several algorithms including the Rank Centrality algorithm, the Matrix Borda algorithm, and the SVM-RankAggregation algorithm succeed in recoverin...
[]
null
71
null
null
[ -0.03371938318014145, -0.019922172650694847, -0.018533745780587196, 0.038590576499700546, 0.032467883080244064, -0.007433244958519936, -0.0014025793643668294, 0.019751543179154396, -0.025587767362594604, -0.051184650510549545, -0.0056409938260912895, -0.007649781182408333, -0.050578754395246...
Stochastic Dual Coordinate Ascent with Adaptive Probabilities
https://proceedings.mlr.press/v37/csiba15.html
[ "Dominik Csiba", "Zheng Qu", "Peter Richtarik" ]
null
null
This paper introduces AdaSDCA: an adaptive variant of stochastic dual coordinate ascent (SDCA) for solving the regularized empirical risk minimization problems. Our modification consists in allowing the method adaptively change the probability distribution over the dual variables throughout the iterative process. AdaSD...
[]
null
72
1502.08053
title_snapshot
[ -0.03395528718829155, -0.0002335495810257271, 0.012925252318382263, 0.05484596639871597, 0.0461856871843338, 0.069139264523983, 0.013277190737426281, -0.02774791046977043, -0.006224606651812792, -0.06502416729927063, -0.006814806256443262, -0.004612376447767019, -0.040597788989543915, 0.00...
Vector-Space Markov Random Fields via Exponential Families
https://proceedings.mlr.press/v37/tansey15.html
[ "Wesley Tansey", "Oscar Hernan Madrid Padilla", "Arun Sai Suggala", "Pradeep Ravikumar" ]
null
null
We present Vector-Space Markov Random Fields (VS-MRFs), a novel class of undirected graphical models where each variable can belong to an arbitrary vector space. VS-MRFs generalize a recent line of work on scalar-valued, uni-parameter exponential family and mixed graphical models, thereby greatly broadening the class o...
[]
null
73
1505.05117
title_snapshot
[ 0.007773821707814932, 0.013850400224328041, 0.0013242056593298912, 0.03966079652309418, 0.05739337578415871, 0.039009369909763336, 0.031002530828118324, 0.035555873066186905, -0.014092196710407734, -0.05612468346953392, 0.015300928615033627, 0.007054619956761599, -0.06318523734807968, -0.0...
JUMP-Means: Small-Variance Asymptotics for Markov Jump Processes
https://proceedings.mlr.press/v37/hugginsa15.html
[ "Jonathan Huggins", "Karthik Narasimhan", "Ardavan Saeedi", "Vikash Mansinghka" ]
null
null
Markov jump processes (MJPs) are used to model a wide range of phenomenon from disease progression to RNA path folding. However, existing methods suffer from a number of shortcomings: degenerate trajectories in the case of ML estimation of parametric models and poor inferential performance in the case of nonparametric ...
[]
null
74
1503.00332
title_snapshot
[ -0.014036798849701881, -0.005128171760588884, -0.032615844160318375, 0.0075546978041529655, 0.03575437143445015, 0.019848506897687912, 0.06630969047546387, 0.004768990911543369, -0.014191655442118645, -0.03990120813250542, 0.027328582480549812, -0.021619398146867752, -0.07009822130203247, ...
Low Rank Approximation using Error Correcting Coding Matrices
https://proceedings.mlr.press/v37/ubaru15.html
[ "Shashanka Ubaru", "Arya Mazumdar", "Yousef Saad" ]
null
null
Low-rank matrix approximation is an integral component of tools such as principal component analysis (PCA), as well as is an important instrument used in applications like web search models, text mining and computer vision, e.g., face recognition. Recently, randomized algorithms were proposed to effectively construct l...
[]
null
75
null
null
[ -0.014506899751722813, -0.02074301242828369, 0.009389970451593399, 0.03002323769032955, 0.03367654234170914, 0.03469366207718849, 0.016439614817500114, -0.00013913576549384743, -0.030306952074170113, -0.040912073105573654, -0.008915175683796406, -0.02213696576654911, -0.06176320090889931, ...
Off-policy Model-based Learning under Unknown Factored Dynamics
https://proceedings.mlr.press/v37/hallak15.html
[ "Assaf Hallak", "Francois Schnitzler", "Timothy Mann", "Shie Mannor" ]
null
null
Off-policy learning in dynamic decision problems is essential for providing strong evidence that a new policy is better than the one in use. But how can we prove superiority without testing the new policy? To answer this question, we introduce the G-SCOPE algorithm that evaluates a new policy based on data generated by...
[]
null
76
null
null
[ -0.005889627151191235, -0.006303878966718912, -0.00034685569698922336, 0.032520782202482224, 0.04951165243983269, 0.005963686853647232, 0.01847943291068077, 0.0006568765384145081, -0.016307469457387924, -0.01882948726415634, -0.005700399633497, 0.033947013318538666, -0.1005665510892868, 0....
Log-Euclidean Metric Learning on Symmetric Positive Definite Manifold with Application to Image Set Classification
https://proceedings.mlr.press/v37/huanga15.html
[ "Zhiwu Huang", "Ruiping Wang", "Shiguang Shan", "Xianqiu Li", "Xilin Chen" ]
null
null
The manifold of Symmetric Positive Definite (SPD) matrices has been successfully used for data representation in image set classification. By endowing the SPD manifold with Log-Euclidean Metric, existing methods typically work on vector-forms of SPD matrix logarithms. This however not only inevitably distorts the geome...
[]
null
77
null
null
[ 0.002317579463124275, -0.007194326724857092, 0.010640406049787998, 0.03180485963821411, 0.02238236740231514, 0.055392250418663025, 0.007484475150704384, -0.010588090866804123, -0.039362870156764984, -0.056133732199668884, -0.027101485058665276, -0.013569752685725689, -0.06442710012197495, ...
Asymmetric Transfer Learning with Deep Gaussian Processes
https://proceedings.mlr.press/v37/kandemir15.html
[ "Melih Kandemir" ]
null
null
We introduce a novel Gaussian process based Bayesian model for asymmetric transfer learning. We adopt a two-layer feed-forward deep Gaussian process as the task learner of source and target domains. The first layer projects the data onto a separate non-linear manifold for each task. We perform knowledge transfer by pro...
[]
null
78
null
null
[ -0.007632976397871971, -0.015822164714336395, -0.015054738149046898, 0.014382544904947281, 0.01821913756430149, 0.016044743359088898, 0.021514661610126495, -0.013171272352337837, 0.013445192947983742, -0.04476243630051613, 0.006672240793704987, 0.025557326152920723, -0.04831632599234581, 0...
Towards a Lower Sample Complexity for Robust One-bit Compressed Sensing
https://proceedings.mlr.press/v37/zhua15.html
[ "Rongda Zhu", "Quanquan Gu" ]
null
null
In this paper, we propose a novel algorithm based on nonconvex sparsity-inducing penalty for one-bit compressed sensing. We prove that our algorithm has a sample complexity of O(s/ε^2) for strong signals, and O(s\log d/ε^2) for weak signals, where s is the number of nonzero entries in the signal vector, d is the signal...
[]
null
79
null
null
[ -0.00788821280002594, -0.029018109664320946, 0.01360184233635664, 0.021167920902371407, 0.06394024938344955, 0.027847478166222572, 0.023317253217101097, -0.013349788263440132, -0.039009373635053635, -0.06976021826267242, 0.016958054155111313, -0.023122666403651237, -0.052277371287345886, 0...
BilBOWA: Fast Bilingual Distributed Representations without Word Alignments
https://proceedings.mlr.press/v37/gouws15.html
[ "Stephan Gouws", "Yoshua Bengio", "Greg Corrado" ]
null
null
We introduce BilBOWA (Bilingual Bag-of-Words without Alignments), a simple and computationally-efficient model for learning bilingual distributed representations of words which can scale to large monolingual datasets and does not require word-aligned parallel training data. Instead it trains directly on monolingual dat...
[]
null
80
1410.2455
title_snapshot
[ -0.012669963762164116, -0.024607524275779724, -0.030395345762372017, 0.019276537001132965, 0.007029722444713116, 0.003906241385266185, 0.013905465602874756, 0.01078860554844141, -0.027817023918032646, -0.02511199563741684, -0.019243814051151276, 0.02353385090827942, -0.07475719600915909, 0...
Multi-view Sparse Co-clustering via Proximal Alternating Linearized Minimization
https://proceedings.mlr.press/v37/sunb15.html
[ "Jiangwen Sun", "Jin Lu", "Tingyang Xu", "Jinbo Bi" ]
null
null
When multiple views of data are available for a set of subjects, co-clustering aims to identify subject clusters that agree across the different views. We explore the problem of co-clustering when the underlying clusters exist in different subspaces of each view. We propose a proximal alternating linearized minimizatio...
[]
null
81
null
null
[ -0.014675025828182697, 0.006896524224430323, 0.02622588537633419, 0.02454620972275734, 0.0338062085211277, 0.027701042592525482, 0.005473659373819828, -0.011948431842029095, -0.031029125675559044, -0.0392141193151474, -0.008907327428460121, -0.02745465189218521, -0.06934407353401184, -0.00...
Cascading Bandits: Learning to Rank in the Cascade Model
https://proceedings.mlr.press/v37/kveton15.html
[ "Branislav Kveton", "Csaba Szepesvari", "Zheng Wen", "Azin Ashkan" ]
null
null
A search engine usually outputs a list of K web pages. The user examines this list, from the first web page to the last, and chooses the first attractive page. This model of user behavior is known as the cascade model. In this paper, we propose cascading bandits, a learning variant of the cascade model where the object...
[]
null
82
1502.02763
title_snapshot
[ -0.018111201003193855, -0.017227306962013245, 0.012727735564112663, 0.05070897564291954, 0.0324237160384655, -0.004272158723324537, 0.018117301166057587, 0.029916593804955482, -0.010928140953183174, -0.025802241638302803, -0.0017937523080036044, 0.0034849168732762337, -0.05829174444079399, ...
Latent Topic Networks: A Versatile Probabilistic Programming Framework for Topic Models
https://proceedings.mlr.press/v37/foulds15.html
[ "James Foulds", "Shachi Kumar", "Lise Getoor" ]
null
null
Topic models have become increasingly prominent text-analytic machine learning tools for research in the social sciences and the humanities. In particular, custom topic models can be developed to answer specific research questions. The design of these models requires a non-trivial amount of effort and expertise, motiva...
[]
null
83
null
null
[ 0.020519046112895012, -0.044917114078998566, -0.03580775111913681, 0.038160715252161026, 0.03992815315723419, 0.002111832844093442, 0.0035228519700467587, 0.017156897112727165, -0.023475946858525276, -0.01087968796491623, -0.010615195147693157, 0.011980501003563404, -0.03793404996395111, -...
Random Coordinate Descent Methods for Minimizing Decomposable Submodular Functions
https://proceedings.mlr.press/v37/ene15.html
[ "Alina Ene", "Huy Nguyen" ]
null
null
Submodular function minimization is a fundamental optimization problem that arises in several applications in machine learning and computer vision. The problem is known to be solvable in polynomial time, but general purpose algorithms have high running times and are unsuitable for large-scale problems. Recent work have...
[]
null
84
1502.02643
title_snapshot
[ -0.01967531070113182, -0.015619122423231602, 0.005654310807585716, 0.050437163561582565, 0.04416859149932861, 0.05000094324350357, -0.005354671739041805, -0.011262096464633942, -0.017130648717284203, -0.04526413977146149, -0.02696613036096096, 0.0025093951262533665, -0.05756129324436188, 0...
Alpha-Beta Divergences Discover Micro and Macro Structures in Data
https://proceedings.mlr.press/v37/narayan15.html
[ "Karthik Narayan", "Ali Punjani", "Pieter Abbeel" ]
null
null
Although recent work in non-linear dimensionality reduction investigates multiple choices of divergence measure during optimization \citeyang2013icml,bunte2012neuro, little work discusses the direct effects that divergence measures have on visualization. We study this relationship, theoretically and through an empirica...
[]
null
85
null
null
[ -0.011854862794280052, -0.013695658184587955, 0.016410598531365395, 0.022414077073335648, 0.023448489606380463, 0.005890657193958759, 0.034531183540821075, -0.006369417067617178, -0.04444735869765282, -0.03450993448495865, 0.005892802029848099, -0.022593189030885696, -0.055965617299079895, ...
Fictitious Self-Play in Extensive-Form Games
https://proceedings.mlr.press/v37/heinrich15.html
[ "Johannes Heinrich", "Marc Lanctot", "David Silver" ]
null
null
Fictitious play is a popular game-theoretic model of learning in games. However, it has received little attention in practical applications to large problems. This paper introduces two variants of fictitious play that are implemented in behavioural strategies of an extensive-form game. The first variant is a full-width...
[]
null
86
null
null
[ -0.05416887626051903, -0.037272579967975616, 0.004243489354848862, 0.012520486488938332, 0.052823860198259354, 0.023845873773097992, -0.00861686933785677, 0.008976630866527557, -0.044114768505096436, -0.0274627897888422, -0.007844977080821991, 0.017351923510432243, -0.06376726925373077, 0....
Counterfactual Risk Minimization: Learning from Logged Bandit Feedback
https://proceedings.mlr.press/v37/swaminathan15.html
[ "Adith Swaminathan", "Thorsten Joachims" ]
null
null
We develop a learning principle and an efficient algorithm for batch learning from logged bandit feedback. This learning setting is ubiquitous in online systems (e.g., ad placement, web search, recommendation), where an algorithm makes a prediction (e.g., ad ranking) for a given input (e.g., query) and observes bandit ...
[]
null
87
1502.02362
title_snapshot
[ 0.011139866895973682, -0.022937403991818428, 0.01614096574485302, 0.0297755878418684, 0.04100670665502548, 0.017199764028191566, 0.03611112758517265, 0.015091491863131523, -0.027260689064860344, -0.03365066647529602, -0.016348866745829582, 0.03859063610434532, -0.07913632690906525, -0.0308...
The Hedge Algorithm on a Continuum
https://proceedings.mlr.press/v37/krichene15.html
[ "Walid Krichene", "Maximilian Balandat", "Claire Tomlin", "Alexandre Bayen" ]
null
null
We consider an online optimization problem on a subset S of R^n (not necessarily convex), in which a decision maker chooses, at each iteration t, a probability distribution x^(t) over S, and seeks to minimize a cumulative expected loss, where each loss is a Lipschitz function revealed at the end of iteration t. Buildin...
[]
null
88
null
null
[ -0.030544035136699677, -0.010427442379295826, 0.023122752085328102, 0.039654940366744995, 0.040636900812387466, 0.03566671535372734, 0.017426013946533203, 0.010511388070881367, 0.007913099601864815, -0.04343250021338463, 0.0005566964391618967, 0.014172661118209362, -0.05454175919294357, -0...
A Linear Dynamical System Model for Text
https://proceedings.mlr.press/v37/belanger15.html
[ "David Belanger", "Sham Kakade" ]
null
null
Low dimensional representations of words allow accurate NLP models to be trained on limited annotated data. While most representations ignore words’ local context, a natural way to induce context-dependent representations is to perform inference in a probabilistic latent-variable sequence model. Given the recent succes...
[]
null
89
1502.04081
title_snapshot
[ -0.02106383629143238, 0.009325144812464714, -0.016828006133437157, 0.05072391778230667, 0.040362510830163956, 0.02685251086950302, 0.03592447564005852, 0.0298914797604084, -0.03172809258103371, -0.016668252646923065, -0.007133139297366142, -0.00046550791012123227, -0.07415815442800522, -0....
Unsupervised Learning of Video Representations using LSTMs
https://proceedings.mlr.press/v37/srivastava15.html
[ "Nitish Srivastava", "Elman Mansimov", "Ruslan Salakhudinov" ]
null
null
We use Long Short Term Memory (LSTM) networks to learn representations of video sequences. Our model uses an encoder LSTM to map an input sequence into a fixed length representation. This representation is decoded using single or multiple decoder LSTMs to perform different tasks, such as reconstructing the input sequen...
[]
null
90
1502.04681
title_snapshot
[ 0.030344583094120026, -0.01980133168399334, -0.03061818890273571, 0.04789680242538452, 0.04299382120370865, 0.026161573827266693, 0.012639554217457771, 0.025695301592350006, -0.02653534524142742, -0.029278524219989777, -0.02451477386057377, 0.0018112119287252426, -0.057982996106147766, 0.0...
Message Passing for Collective Graphical Models
https://proceedings.mlr.press/v37/sunc15.html
[ "Tao Sun", "Dan Sheldon", "Akshat Kumar" ]
null
null
Collective graphical models (CGMs) are a formalism for inference and learning about a population of independent and identically distributed individuals when only noisy aggregate data are available. We highlight a close connection between approximate MAP inference in CGMs and marginal inference in standard graphical mod...
[]
null
91
null
null
[ -0.005039644427597523, -0.005973909515887499, -0.012316501699388027, 0.02918383665382862, 0.036888498812913895, 0.017170676961541176, 0.062312282621860504, 0.027869651094079018, -0.02744213677942753, -0.0515245757997036, 0.020975790917873383, -0.010698542930185795, -0.09876050055027008, 0....
DP-space: Bayesian Nonparametric Subspace Clustering with Small-variance Asymptotics
https://proceedings.mlr.press/v37/wanga15.html
[ "Yining Wang", "Jun Zhu" ]
null
null
Subspace clustering separates data points approximately lying on union of affine subspaces into several clusters. This paper presents a novel nonparametric Bayesian subspace clustering model that infers both the number of subspaces and the dimension of each subspace from the observed data. Though the posterior inferenc...
[]
null
92
null
null
[ -0.01707262359559536, 0.0031342541333287954, -0.0027213844005018473, 0.04231387749314308, 0.034493446350097656, 0.049776554107666016, 0.025677770376205444, -0.021665504202246666, 0.0003639347560238093, -0.04307347908616066, -0.015905704349279404, -0.013064282946288586, -0.07886876165866852, ...
HawkesTopic: A Joint Model for Network Inference and Topic Modeling from Text-Based Cascades
https://proceedings.mlr.press/v37/he15.html
[ "Xinran He", "Theodoros Rekatsinas", "James Foulds", "Lise Getoor", "Yan Liu" ]
null
null
Understanding the diffusion of information in social network and social media requires modeling the text diffusion process. In this work, we develop the HawkesTopic model (HTM) for analyzing text-based cascades, such as "retweeting a post" or "publishing a follow-up blog post". HTM combines Hawkes processes and topic m...
[]
null
93
null
null
[ 0.0019491236889734864, -0.007947216741740704, -0.013915727846324444, 0.04639117419719696, 0.031720153987407684, -0.01922946609556675, 0.020235707983374596, 0.026244185864925385, -0.014179336838424206, -0.023052558302879333, 0.025706270709633827, 0.018235426396131516, -0.026445189490914345, ...
MADE: Masked Autoencoder for Distribution Estimation
https://proceedings.mlr.press/v37/germain15.html
[ "Mathieu Germain", "Karol Gregor", "Iain Murray", "Hugo Larochelle" ]
null
null
There has been a lot of recent interest in designing neural network models to estimate a distribution from a set of examples. We introduce a simple modification for autoencoder neural networks that yields powerful generative models. Our method masks the autoencoder’s parameters to respect autoregressive constraints: ea...
[]
null
94
1502.03509
title_snapshot
[ 0.006977663841098547, -0.0029404445085674524, -0.013316848315298557, 0.04739822447299957, 0.04793369770050049, 0.09806393086910248, 0.018313206732273102, -0.010433479212224483, -0.041714318096637726, -0.0488198883831501, -0.009005670435726643, -0.009037341922521591, -0.06222881004214287, 0...
An Online Learning Algorithm for Bilinear Models
https://proceedings.mlr.press/v37/wua15.html
[ "Yuanbin Wu", "Shiliang Sun" ]
null
null
We investigate the bilinear model, which is a matrix form linear model with the rank 1 constraint. A new online learning algorithm is proposed to train the model parameters. Our algorithm runs in the manner of online mirror descent, and gradients are computed by the power iteration. To analyze it, we give a new second ...
[]
null
95
null
null
[ -0.04219330474734306, -0.004019685555249453, 0.0029561093542724848, -0.0062249149195849895, 0.015273626893758774, 0.013974909670650959, 0.0178156029433012, 0.006213102024048567, -0.00602481234818697, -0.02828036993741989, -0.01240713894367218, 0.0067977593280375, -0.07416195422410965, -0.0...
Adaptive Belief Propagation
https://proceedings.mlr.press/v37/papachristoudis15.html
[ "Georgios Papachristoudis", "John Fisher" ]
null
null
Graphical models are widely used in inference problems. In practice, one may construct a single large-scale model to explain a phenomenon of interest, which may be utilized in a variety of settings. The latent variables of interest, which can differ in each setting, may only represent a small subset of all variables. T...
[]
null
96
null
null
[ -0.00564615847542882, -0.01599595695734024, 0.002752584870904684, 0.03854810819029808, 0.05138194561004639, 0.034273382276296616, 0.044299859553575516, 0.009607076644897461, -0.02178080938756466, -0.061453625559806824, -0.005481843836605549, 0.027526281774044037, -0.07438292354345322, 0.00...
Large-scale log-determinant computation through stochastic Chebyshev expansions
https://proceedings.mlr.press/v37/hana15.html
[ "Insu Han", "Dmitry Malioutov", "Jinwoo Shin" ]
null
null
Logarithms of determinants of large positive definite matrices appear ubiquitously in machine learning applications including Gaussian graphical and Gaussian process models, partition functions of discrete graphical models, minimum-volume ellipsoids and metric and kernel learning. Log-determinant computation involves t...
[]
null
97
1503.06394
title_snapshot
[ -0.0195649191737175, 0.01757752150297165, -0.00016604563279543072, -0.01330493576824665, 0.029139025136828423, 0.027435818687081337, 0.02467343583703041, -0.008670207113027573, -0.021904723718762398, -0.02113465592265129, -0.00015194856678135693, -0.019888179376721382, -0.08121654391288757, ...
Differentially Private Bayesian Optimization
https://proceedings.mlr.press/v37/kusnera15.html
[ "Matt Kusner", "Jacob Gardner", "Roman Garnett", "Kilian Weinberger" ]
null
null
Bayesian optimization is a powerful tool for fine-tuning the hyper-parameters of a wide variety of machine learning models. The success of machine learning has led practitioners in diverse real-world settings to learn classifiers for practical problems. As machine learning becomes commonplace, Bayesian optimization bec...
[]
null
98
1501.04080
title_snapshot
[ -0.011665621772408485, 0.023729441687464714, 0.004827587399631739, 0.045384153723716736, 0.04739135503768921, 0.037548791617155075, 0.05407719686627388, -0.0514984130859375, 0.00014426409325096756, -0.026046650484204292, -0.0071332077495753765, 0.018593573942780495, -0.038998402655124664, ...
A Nearly-Linear Time Framework for Graph-Structured Sparsity
https://proceedings.mlr.press/v37/hegde15.html
[ "Chinmay Hegde", "Piotr Indyk", "Ludwig Schmidt" ]
null
null
We introduce a framework for sparsity structures defined via graphs. Our approach is flexible and generalizes several previously studied sparsity models. Moreover, we provide efficient projection algorithms for our sparsity model that run in nearly-linear time. In the context of sparse recovery, we show that our framew...
[]
null
99
null
null
[ 0.00922954361885786, -0.012383962981402874, 0.03344929590821266, 0.02204308472573757, 0.042499881237745285, 0.03741796314716339, 0.00938569288700819, 0.005730036646127701, -0.02887294813990593, -0.05816573649644852, 0.028608767315745354, -0.027896223589777946, -0.05842633917927742, -0.0015...
Support Matrix Machines
https://proceedings.mlr.press/v37/luo15.html
[ "Luo Luo", "Yubo Xie", "Zhihua Zhang", "Wu-Jun Li" ]
null
null
In many classification problems such as electroencephalogram (EEG) classification and image classification, the input features are naturally represented as matrices rather than vectors or scalars. In general, the structure information of the original feature matrix is useful and informative for data analysis tasks such...
[]
null
100
null
null
[ -0.01855023391544819, -0.011951807886362076, 0.007930619642138481, -0.01737501285970211, 0.010441729798913002, 0.03992313891649246, 0.02542450651526451, -0.027541259303689003, -0.05327053368091583, -0.04283478111028671, -0.007084163837134838, 0.012278069742023945, -0.07473567128181458, 0.0...