Aggregation by Exponential Weighting, Sharp PAC-Bayesian Bounds and Sparsity

Arnak S. Dalalyan, Alexandre B. Tsybakov

MLJ 2008 pp. 39-61

doi:10.1007/S10994-008-5051-0 /mlj/2008/dalalyan2008mlj-aggregation/

Abstract

We study the problem of aggregation under the squared loss in the model of regression with deterministic design. We obtain sharp PAC-Bayesian risk bounds for aggregates defined via exponential weights, under general assumptions on the distribution of errors and on the functions to aggregate. We then apply these results to derive sparsity oracle inequalities.

PDF MLJ Semantic Scholar

Cite

Text

Dalalyan and Tsybakov. "Aggregation by Exponential Weighting, Sharp PAC-Bayesian Bounds and Sparsity." Machine Learning, 2008. doi:10.1007/S10994-008-5051-0

Markdown

[Dalalyan and Tsybakov. "Aggregation by Exponential Weighting, Sharp PAC-Bayesian Bounds and Sparsity." Machine Learning, 2008.](https://mlanthology.org/mlj/2008/dalalyan2008mlj-aggregation/) doi:10.1007/S10994-008-5051-0

BibTeX

@article{dalalyan2008mlj-aggregation,
  title     = {{Aggregation by Exponential Weighting, Sharp PAC-Bayesian Bounds and Sparsity}},
  author    = {Dalalyan, Arnak S. and Tsybakov, Alexandre B.},
  journal   = {Machine Learning},
  year      = {2008},
  pages     = {39-61},
  doi       = {10.1007/S10994-008-5051-0},
  volume    = {72},
  url       = {https://mlanthology.org/mlj/2008/dalalyan2008mlj-aggregation/}
}