2009

Learning Exponential Families in High-Dimensions: Strong Convexity and Sparsity

Kakade, Sham M., Shamir, Ohad, Sridharan, Karthik et al.

Understand

The versatility of exponential families, along with their attendant convexity properties, make them a popular and effective statistical model.

  • A central issue is learning these models in high-dimensions, such as when there is some sparsity pattern of the optimal parameter.
  • This work characterizes a certain strong convexity property of general exponential families, which allow their generalization ability to be quantified.
  • In particular, we show how this property can be used to analyze generic exponential families under L_1 regularization.

Built on

  • The Theory of Probabilities

    S. Bernstein · 1946

    Earlier work this paper cites.

  • Fundamentals of Statistical Exponential Families

    Lawrence D. Brown · 1986

    Earlier work this paper cites.

  • Asymptotic behavior of likelihood methods for exponential families when the number of parameters tends to infinity

    S. Portnoy · 1988

    Earlier work this paper cites.

  • An Introduction to Generalized Linear Models

    A.J. Dobson · 1990

    Earlier work this paper cites.

  • Asymptotic normality of posterior distributions for exponential families when the number of parameters tends to infinity

    Subhashis Ghosal · 2000

    Earlier work this paper cites.

Similar

  • A Primer of Real Analytic Functions

    Steven G. Krantz and Harold R. Parks · 2002

    Cited alongside, same era.

  • Convex Optimization

    S. Boyd and L. Vandenberghe · 2004

    Cited alongside, same era.

  • On model selection consistency of lasso

    Original

    P. Zhao and B. Yu · 2006

    Cited alongside, same era.

  • The Dantzig selector: Statistical estimation when p p is much larger than n n

    Emmanuel Candes and Terence Tao · 2007

    Cited alongside, same era.

  • High-dimensional ising model selection using l1-regularized logistic regression

    P. Ravikumar, M. J. Wainwright, and J. Lafferty

    Cited in the paper.

  • High-dimensional covariance estimation by minimizing l1-penalized log-determinant divergence

    Original

    Pradeep Ravikumar, Martin J Wainwright, Garvesh Raskutti, and Bin Yu

    Cited in the paper.

Then

  • Simultaneous analysis of Lasso and Dantzig selector

    Peter J. Bickel, Ya’acov Ritov, and Alexandre B. Tsybakov · 2008

    Later among the works it cites.

  • Adaptive forward-backward greedy algorithm for sparse learning with linear models

    T. Zhang · 2008

    Later among the works it cites.

  • Lasso-type recovery of sparse representations for high-dimensional data

    Nicolai Meinshausen and Bin Yu · 2009

    Closest in time.

  • Trading accuracy for sparsity

    S. Shalev-Shwartz, N. Srebro, and T. Zhang · 2009

    Closest in time.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…