Statistical Machine Learning Reading Group

Carnegie Mellon University

Room: 7501 Gates-Hillman Center

Time: noon-1:30 pm Wednesday


SCHEDULE:

Jan 14 High Dimensional Structure Learning of Ising Models on Sparse Random Graphs (arXiv)
Authors: Animashree Anandkumar, Vincent Tan, Alan Willsky
Presenter: Divyanshu Vats

Jan 21 Detection of an anomalous cluster in a network (arXiv)
Authors: Ery Arias-Castro, Emmanuel J. Candes, Arnaud Durand
Presenter: James Sharpnack

Jan 26 The Sample Complexity of Dictionary Learning (arXiv)
Authors: Daniel Vainsencher, Shie Mannor, Alfred M. Bruckstein
Presenter: Min Xu

Feb 2 Concentration inequalities of the cross-validation estimate for stable predictors (arXiv)
Authors: Matthieu Cornec
Presenter: Martin Azizyan

Feb 9 Stability Bounds for Stationary phi-mixing and beta-mixing Processes (here)
Authors: Mehryar Mohri, Afshin Rostamizadeh
Presenter: Dan McDonald

Feb 16 Nuclear norm penalization and optimal rates for noisy low rank matrix completion (arXiv)
Authors: Vladimir Koltchinskii, Alexandre B. Tsybakov, Karim Lounici
Presenter: Akshay Krishnamurthy

Feb 23 Information theoretic model validation for clustering (arXiv)
Authors: Joachim M. Buhmann
Presenter: Larry Wasserman

Mar 4 (12:30 in GHC 4211) Noisy matrix decomposition via convex relaxation: Optimal rates in high dimensions(arXiv)
Authors: Alekh Agarwal
Presenter: Alekh Agarwal

Mar 16 Dynamics of Bayesian updating with dependent data and misspecified models (Euclid)
Authors: Cosma Shalizi
Presenter: Cosma Shalizi

Mar 23 Online Learning (here and here)
Authors: Alex Rakhlin et al.
Presenter: Aaditya Ramdas

Mar 30 High-dimensional analysis of semidefinite relaxations for sparse principal components (Euclid)
Authors: Arash A. Amini and Martin J. Wainwright
Presenter: Mladen Kolar

Apr 20 Empirical risk minimization in inverse problems. (Euclid)
Authors: Jussi Klemela and Enno Mammen
Presenter: Darren Homrighausen


Reading list requests/suggestions:
  • Concentration inequalities of the cross-validation estimator for Empirical Risk Minimiser by Matthieu Cornec arXiv
  • Rademacher Complexities and Bounding the Excess Risk in Active Learning by Vladimir Koltchinskii pdf
  • PAC-Bayesian Analysis of Co-clustering and Beyond by Y. Seldin and N. Tishby pdf
  • Asymptotic Equivalence of Bayes Cross Validation and Widely Applicable Information Criterion in Singular Learning Theory by Sumio Watanabe pdf