Statistical Machine Learning Reading Group

SML Student Workshop Summer 2016

Date: Wednesday, June 1
Location: 335 external link: West Hall
Organizing Faculty: Jake Abernethy, Laura Balzano, Long Nguyen, Clay Scott, Ambuj Tewari

external link: Full schedule with abstracts

9:15 Welcome and Coffee
 Session 1: Convergence and Consistency
9:30Nhat HoSingularity structures and parameter estimation in finite skew normal mixtures.
9:45Julian Katz-SamuelsA Mutual Contamination Analysis of Mixed Membership and Partial Label Models
10:00Kevin MoonImproving Convergence of Divergence Functional Ensemble Estimators
10:15Hossein KeshavarzFixed domain asymptotic analysis of the inversion-free estimation for Gaussian process models
10:30 Break
 Session 2: Subspace Models
10:45David HongA weighted PCA method for subspace estimation from heterogeneous data
11:00Kibok LeeTowards Understanding the Invertibility of Convolutional Neural Networks
11:15John LiporLeveraging Union of Subspace Structure to Improve Constrained Clustering
11:30 Lunch

Winter 2016
Time: Mondays 3-4
Location: BBB 2733 (on north campus)

DatePresenterPaperAuthorsAppears In
2/8HosseinStochastic approximation of score functions for Gaussian processesM.L. Stein, J. Chen, and M. AnitescuThe Annals of Applied Statistics 7.2 (2013) external link: http://arxiv.org/pdf/1312.2687.pdf
2/15Aniket DeshmukhPermutational Rademacher ComplexityIlya Tolstikhin, Nikita Zhivotovskiy, and Gilles BlanchardAlgorithmic Learning Theory. Springer International Publishing, 2015 (external link: http://arxiv.org/abs/1505.02910)
3/14Julianℓ1-regularized Neural Networks are Improperly Learnable in Polynomial TimeYuchen Zhang, Jason D. Lee, Michael I. Jordanexternal link: http://arxiv.org/abs/1510.03528
3/21Aniket DeshmukhFinite-Time Analysis of Kernelised Contextual BanditsMichal Valko, Nathan Korda, Remi Munos, Ilias Flaounas, Nello Cristianiniexternal link: http://arxiv.org/abs/1309.6869
3/28Daniel LeJeuneOptimal Rates for Random Fourier FeaturesBharath K. Sriperumbudur, Zoltan SzaboNIPS 2015 (external link: http://arxiv.org/abs/1506.02155)
4/18HosseinLocal likelihood estimation for nonstationary random fieldsE.B. Anderes, and M.L. SteinJournal of Multivariate Analysis (2011) external link: http://www.stat.ucdavis.edu/~anderes/papers/LocLikeWStein.pdf

Previous Semesters

Fall 2015
Time: Wednesdays, 12:15-1:15
Location: EECS 2311 (on north campus)

DatePresenterPaperAuthorsAppears In
Sept 23ClayThe Geometry of Kernelized Spectral ClusteringSchiebinger, Wainwright, Yuexternal link: Annals of Statistics, 2015.
Sept 30HosseinCovariance functions for mean square differentiable processes on spheresJ. Guinness and M. Fuentesexternal link: http://www.stat.ncsu.edu/information/library/papers/mimeo2652_Guinness.pdf 
Oct 7HarishConsistent Procedures for Cluster Tree Estimation and PruningChaudhuri, Dasgupta, Kpotufe and Von Luxburgexternal link: IEEE Info theory, 2014
Oct 14EfrenLearning with Algebraic Invariances and the Invariant Kernel TrickKirály, Ziehe, Müllerexternal link: http://arxiv.org/abs/1411.7817
Oct 21RobNonparametric Estimation of Multi-View Latent Variable ModelsLe Song, Animashree Anandkumar, Bo Dai, Bo XieNIPS 2014
Oct 28JulianRates of Convergence for Nearest Neighbor ClassificationChaudhuri, Dasguptaexternal link: http://cseweb.ucsd.edu/~dasgupta/papers/nn-rates.pdf 
Nov 4Nhat HoMixture models with a prior on the number of componentsJeff Miller, Matthew HarrisonUnder revision, Journal of the American Statistical Association, external link: https://stat.duke.edu/~jwm40/publications/MFM.pdf
Nov 11Akshay KrishnamurthyEfficient Contextual Semibanditswith Alekh Agarwal and Miro Dudikexternal link: http://arxiv.org/pdf/1502.05890.pdf I will describe a variant of the contextual bandit problem, known as contextual semibandits, where in each round the learner receives a context, plays a sequence of actions, observes a feature for each of the played actions, and observes reward that is linearly related to those features. This setting is motivated by problems in personalized search and recommendation, where many common performance metrics are linearly related to observable document-specific click information. I will describe two algorithms for this problem, one for the case where the linear transformation is known and one for the case where it is unknown. Both algorithms have low regret guarantees and can be efficiently implemented with an appropriate optimization oracle. I will present some preliminary empirical findings on these algorithms and discuss some new ongoing progress in this direction.
Nov 18AniketMultiple Operator-valued Kernel LearningHachem Kadri, Alain Rakotomamonjy, Philippe Preux, Francis R. BachNIPS 2012 - external link: https://papers.nips.cc/paper/4653-multiple-operator-valued-kernel-learning
Dec 2Kam Chung WongOnline Time Series Prediction with Missing DataOren Anava, Elad Hazan, Assaf Zeeviexternal link: http://jmlr.org/proceedings/papers/v37/anava15.html
Dec 9Mikhail Y.K-means Clustering via Principal Component AnalysisChris Ding, Xiaofeng Heexternal link: ICML 2004

SML Student Workshop Summer 2015

Date: Wednesday, June 10
Location: 335 external link: West Hall
Organizing Faculty: Jake Abernethy, Laura Balzano, Long Nguyen, Clay Scott, Ambuj Tewari

external link: Full schedule with abstracts

8:45 Welcome and Coffee
9:00FacultyIntroductory Remarks
 Session 1: Convergence and Consistency
9:05Nhat HoWeak identifiability and optimal convergence rate of mixing measures in over-fitted Gaussian mixture models
9:20Efren Cruz CortesConsistency of a Fixed Bandwidth Kernel Density Estimator
9:35Hossein KeshavarzOn the consistency of the inversion-free estimation of Gaussian random fields for irregularly spaced spatial data
9:50Dejiao ZhangGlobal Convergence of the GROUSE algorithm for subspace estimation from undersampled data
10:05 Break + Q&A
 Session 2: Sparsity and Kernels
10:20David HongAdaptive Dictionary Learning with Training Images for Image Formation
10:35Aniket DeshmukhKernel Approximation for Transfer Learning
10:50Pengyu XiaoGrassmannian Online Sparse Subspace Estimation
11:05 Break + Q&A
 Session 3: Mixture Models, Meta Learning, and Link Prediction
11:20Robert VandermeulenSome Results on the Identifiability of Nonparametric Finite Mixture Models with Grouped Samples.
11:35Bopeng LiHandling Class Imbalance in Link Prediction using Learning to Rank Techniques
11:50Kevin MoonMeta learning of bounds on the Bayes classifier error
12:05 Lunch

Winter 2015
Time: Wednesdays, 3:00-4:00
Location: 438 West Hall (on central campus)

DatePresenterPaperAuthorsAppears In
1/21Wendy ShangInvariant Scattering Convolution NetworksJ. Bruna and S. Mallat 
1/28Laura BalzanoGlobal Convergence of Stochastic Gradient Descent for some Nonconvex Matrix ProblemsChris De Sa, Kunle Olukotun, Chris Rearxiv external link: http://arxiv.org/pdf/1411.1134v2.pdf
2/11RobTensor Decompositions for Learning Latent Variable ModelsAnandkumar et al.JMLR 2014
2/18Mikhail YStreaming Variational BayesBroderick et al.external link: NIPS 2013
2/25SougataFrom Bandits to Experts: On the Value of Side-ObservationsShie Mannor and Ohad Shamirexternal link: http://papers.nips.cc/paper/4366-from-bandits-to-experts-on-the-value-of-side-observations
3/11external link: Igor Prunster, University of Torino/ U Texas-AustinAre Gibbs-type priors the most natural generalization of the Dirichlet process?Joint seminar with Stats Dept 411 West Hall (Note: non-standard location) 3-4pm
3/18Nhat HoVariable selection for k-means quantizationClément LevrardSubmit to the Annals of Statistics, 2015 external link: http://arxiv.org/pdf/1406.3334v1.pdf
3/25HosseinOn the consistent separation of scale and variance for Gaussian random fieldsE. AnderesAnnals of Statistics, 2010, external link: http://projecteuclid.org/euclid.aos/1266586617
4/1HosseinEstimating deformations of isotropic Gaussian random fields on the planeE. Anderes and S. ChatterjeeAnnals of Statistics, 2009, external link: http://projecteuclid.org/euclid.aos/1247663757
4/8Bala Rajaratnam, Standford Universityexternal link: Principled and Scalable Methods for Extracting multivariate dependencies in Big Data  EECS 1200 {red}
4/15Nhat HoInference for Mixtures of Symmetric DistributionsDavid R.Hunter, Shaoli Wang, and Thomas P. HettmanspergerAnnals of Statistics, 2006, external link: http://arxiv.org/pdf/0708.0499.pdf

Fall 2014
Time: Wednesdays, 2:30-4:00
Location: 4246 Randall (on central campus)

DatePresenterPaperAuthorsAppears In
9/10MatusLoss minimization and parameter estimation with heavy tailsDaniel Hsu and Sivan SabatoICML 2014, external link: http://arxiv.org/abs/1307.1827
9/24Nick A.Clustering by fast search and find of density peaksAlex Rodriguez and Alessandro LaioScience June 2014, external link: http://www.sciencemag.org/content/344/6191/1492.full.pdf
10/1JJAlgorithmic connections between active learning and stochastic convex optimizationAaditya Ramdas and Aarti Singhexternal link: http://www.cs.cmu.edu/~aarti/pubs/ALT13_ARamdas.pdf
10/8SougataTop Rank Optimization in Linear TimeNan Li, Rong Jin, Zhi-Hua Zhouexternal link: http://nips.cc/Conferences/2014/Program/event.php?ID=4797
10/15Jun GRank-One Matrix Pursuit for Matrix CompletionZheng Wang,Ming-Jun Lai,Zhaosong Lu,Wei Fan,Hasan Davulcu,Jieping Yeexternal link: http://jmlr.org/proceedings/papers/v32/wanga14.pdf#9
10/22external link: Shivani Agarwalexternal link: Statistical Learning in Complex Prediction Spaces: What Do We Know? 4464 East Hall (Note: non-standard location)
10/29HosseinAsymptotic analysis of the role of spatial sampling for covariance parameter estimation of Gaussian processesFrançois BachocaJournal of Multivariate Analysis 2014, external link: http://www.sciencedirect.com/science/article/pii/S0047259X13002571
11/5ENCCDensity Estimation in Infinite Dimensional Exponential FamiliesSriperumbudur et al.external link: arxiv
11/12Nhat HoRates of convergence for the posterior distributions of mixtures of betas and adaptive nonparametric estimation of the densityJudith RousseauAnnals of Statistics 2010, external link: http://projecteuclid.org/download/pdfview_1/euclid.aos/1262271612
11/19Mikhail YHierarchical Dirichlet Scaling ProcessDongwoo Kim, Alice Ohexternal link: ICML 2014
12/3Kamchung WongOn Iterative Hard Thresholding Methods for High-dimensional M-EstimationPrateek Jain, Ambuj Tewari, Purushottam KarNIPS2014 external link: external link: http://arxiv.org/pdf/1410.5137.pdf
12/10Bopeng LiCovariate Assisted Spectral ClusteringNorbert Binkiewicz, Joshua T. Vogelstein, and Karl Roheexternal link: external link: http://arxiv.org/pdf/1411.2158v1.pdf

SML Student Workshop
Summer 2014
Date: Wednesday, June 25
Location: 335 external link: West Hall
Organizing Faculty: Jake Abernethy, Laura Balzano, Long Nguyen, Clay Scott, Ambuj Tewari

TimePresenterTitleAppears In (if published)Additional comments or notes
  Session 1: Identifiability, Estimation, and Convergence  
9:05Robert VandermeulenEstimation of a Measure Over Densities from Pairs of Samples external link: abstract
9:20Kevin MoonEnsemble estimation of multivariate f-divergenceISIT 2014external link: abstract
9:35Nhat HoIdentifiability and Convergence Rate of Parameter Estimators in Finite Mixture ModelsTBDexternal link: abstract
9:50Pradeep RanganathanLocally-weighted Homographies for Calibration of Imaging Systems external link: abstract
10:05-10:20 15 minute Q&A time and break  
  Session 2: Health and Markets  
10:20Takanori WatanabeDisease Prediction based on Functional Connectomes using a Scalable and Spatially-Informed Support Vector MachineNeuroImageexternal link: abstract
10:35Dae JungComputerized Analysis of the 12-Lead Electrocardiogram to Identify Ventricular Tachycardia Exit SitesHeart Rhythmexternal link: abstract
10:50Huitian LeiOnline Contextual Bandits with Stochastic PolicyTBDexternal link: abstract
11:05Sindhu KuttyExponential Family Prediction MarketsEC 2014external link: abstract
11:20-11:35 15 minute Q&A time and break  
11:35-11:55FacultyDiscussion Panel  
11:55-1:20 Lunch break At external link: Silvio's
  Session 3: Scalability, High-dimensional Statistics  
1:20Efren CruzSparse Approximation of a Kernel MeanICASSP 2014external link: abstract
1:35Sougata ChaudhuriPerceptron-like Algorithms and Generalization Bounds in Learning to RankAccepted for presentation at IISAexternal link: abstract
1:50John LiporRobust Blind Calibration via Total Least SquaresICASSP 2014external link: abstract
2:05Kam Chung WongEstimation in High-dimensional Vector Autoregressive Models with Noisy DataTBDexternal link: abstract
2:20Yiwei ZhangHigh Dimensional Covariance Matrix Estimation via the Barra Model external link: abstract
2:35-2:50 15 minutes Q&A time and closing remarks  

Winter 2014
Time: Tuesdays, 12-1
Location: EECS 2311 except on Jan 21 (Beyster 2733) and Feb 18 (Beyster 3725)

DatePresenterPaperAuthorsAppears In
1/21Clayexternal link: Online learning with kernelsJ. Kivinen, A. J Smola, and R. C WilliamsonTrans. Sig. Proc. 2010
2/04Ambujexternal link: Learning from CrowdsRaykar et al.JMLR 2010
2/11NickMystery paper. Topic: Cluster and Manifold Regularization under review
2/18Prateek Jain (MSR India)external link: Provable Alternating Minimization methods for Non-convex OptimizationJoint work with several co-authorsNon-standard location: 3725 BBB
2/25Robertexternal link: A note on Fermat's problemHarold KuhnMathematical Programming 4 (1973)
3/11LongA paper under review. Topic: optimal transport  
3/18Kamexternal link: Estimation in High-dimensional Vector Autoregressive ModelsSumanta Basu and George MichailidisUnder review
3/25Efren BB Cruzexternal link: New Perspectives on k-support and Cluster NormsA. McDonald, M. Pontil and D. StamosarXiv (2014)
4/01Takexternal link: Randomized Nonlinear Component AnalysisD. Lopez-Paz, S. Sra, A. Smola, Z. Ghahramani, B. SchölkopfarXiv (2014)
4/08Nhatexternal link: Consistency of a recursive estimate of mixing distributionSurya T.Todkar, Ryan Martin, and Jayanta K.GhoshAnnals of Statistics, 2009
4/15Hosseinexternal link: A new covariance inequality and applications J. Dedecker and P. DoukhanStochastic Processes and their Applications, 2003

Fall 2013
Time: Wednesdays, 11-12
Location: EECS 2311

DatePresenterPaperAuthorsAppears In
9/11Hosseinexternal link: Detection of an anomalous cluster in a networkE. Arias-Castro, E.J. Candès and A. DurandAnnals of Statistics, 2011
9/18Lauraexternal link: Rank Aggregation via Nuclear Norm MinimizationDavid F. Gleich, Lek-Heng LimACM SIGKDD 2011
9/25Takexternal link: Regularized M-estimators with nonconvexity: Statistical and algorithmic theory for local optimaPo-Ling Loh, Martin J. WainwrightarXiv
10/2Sougataexternal link: Classification Calibration Dimension for General Multiclass LossesHarish Ramaswamy, Shivani AgarwalNIPS 2012
10/9Clayexternal link: Sparse coding for multitask and transfer learningAndreas Maurer, Massi Pontil, Bernardino Romera-ParedesICML 2013
10/16Nickexternal link: Canonical Coordinates are the Right Coordinates for Low-Rank Gauss-Gauss Detection and EstimationAli Pezeshki, Louis L. Scharf, Johnk Thomas, Barry D. Van VeenTrans. on Sig. Proc. 2006
10/23Ambujexternal link: Online Learning for Time Series Prediction external link: Full version with proofsO. Anava, E. Hazan, S. Mannor, O. ShamirCOLT 2013
10/30Harishexternal link: Agnostic Active LearningM.F. Balcan, A. Beygelzeimer and J. Langford.ICML 2006
11/6Clayexternal link: Decontamination of Mutually Contaminated ModelsGilles Blanchard and Clayton ScottAISTATS 2014
11/13Efren Cruz Cortesexternal link: Sampling Methods for the Nystrom MethodS. Kumar, M. Mohri, A. TalwalkarJMLR 2012
11/20Nhat Hoexternal link: Convergence of Latent Mixing Measures in Finite and Infinite Mixture Models XuanLong NguyenAnnals of Statistics 2013
12/4Longexternal link: Borrowing strength in hierarchical Bayes: convergence of the Dirichlet base measureX. NguyenArxiv
12/11Robexternal link: RATES OF STRONG UNIFORM CONSISTENCY FOR MULTIVARIATE KERNEL DENSITY ESTIMATORSEvarist Gine and Armelle GuillouAnn. Inst. Henri Poincaré (B)

Link to external link: previous semesters

Recently changed pages

Changes since Aug 22, 2018 6:17 PM