AISTATS 2019poster7 citations

Error bounds for sparse classifiers in high-dimensions

Antoine Dedieu

Abstract

We prove an L2 recovery bound for a family of sparse estimators defined as minimizers of some empirical loss functions – which include hinge loss and logistic loss. More precisely, we achieve an upper-bound for coefficients estimation scaling as $(k\ast/n)\log(p/k\ast)$: n,p is the size of the design matrix and k* the dimension of the theoretical loss minimizer. This is done under standard assumptions, for which we derive stronger versions of a cone condition and a restricted strong convexity. Our bound holds with high probability and in expectation and applies to an L1-regularized estimator and to a recently introduced Slope estimator, which we generalize for classification problems. Slope presents the advantage of adapting to unknown sparsity. Thus, we propose a tractable proximal algorithm to compute it and assess its empirical performance. Our results match the best existing bounds for classification and regression problems.

BibTeX
@InProceedings{pmlr-v89-dedieu19a,
  title = 	 {Error bounds for sparse classifiers in high-dimensions},
  author =       {Dedieu, Antoine},
  booktitle = 	 {Proceedings of the Twenty-Second International Conference on Artificial Intelligence and Statistics},
  pages = 	 {48--56},
  year = 	 {2019},
  editor = 	 {Chaudhuri, Kamalika and Sugiyama, Masashi},
  volume = 	 {89},
  series = 	 {Proceedings of Machine Learning Research},
  month = 	 {16--18 Apr},
  publisher =    {PMLR},
  pdf = 	 {http://proceedings.mlr.press/v89/dedieu19a/dedieu19a.pdf},
  url = 	 {https://proceedings.mlr.press/v89/dedieu19a.html},
  abstract = 	 {We prove an L2 recovery bound for a family of sparse estimators defined as minimizers of some empirical loss functions – which include hinge loss and logistic loss. More precisely, we achieve an upper-bound for coefficients estimation scaling as $(k\ast/n)\log(p/k\ast)$: n,p is the size of the design matrix and k* the dimension of the theoretical loss minimizer. This is done under standard assumptions, for which we derive stronger versions of a cone condition and a restricted strong convexity. Our bound holds with high probability and in expectation and applies to an L1-regularized estimator and to a recently introduced Slope estimator, which we generalize for classification problems.  Slope presents the advantage of adapting to unknown sparsity. Thus, we propose a tractable proximal algorithm to compute it and assess its empirical performance. Our results match the best existing bounds for classification and regression problems.}
}
Error bounds for sparse classifiers in high-dimensions · AISTATS 2019