Screening Data Points in Empirical Risk Minimization via Ellipsoidal Regions and Safe Loss Functions
Abstract
We design simple screening tests to automatically discard data samples in empirical risk minimization withoutlosing optimization guarantees. We derive loss functions that produce dual objectives with a sparse solution. We also show how to regularize convex losses to ensure such a dual sparsity-inducing property, andpropose a general method to design screening tests for classification or regression based on ellipsoidal approximations of the optimal set. In addition to producing computational gains, our approach also allows us to compress a dataset into a subset of representative points.
Cite
Text
Mialon et al. "Screening Data Points in Empirical Risk Minimization via Ellipsoidal Regions and Safe Loss Functions." Artificial Intelligence and Statistics, 2020.Markdown
[Mialon et al. "Screening Data Points in Empirical Risk Minimization via Ellipsoidal Regions and Safe Loss Functions." Artificial Intelligence and Statistics, 2020.](https://mlanthology.org/aistats/2020/mialon2020aistats-screening/)BibTeX
@inproceedings{mialon2020aistats-screening,
title = {{Screening Data Points in Empirical Risk Minimization via Ellipsoidal Regions and Safe Loss Functions}},
author = {Mialon, Grégoire and Mairal, Julien and d’Aspremont, Alexandre},
booktitle = {Artificial Intelligence and Statistics},
year = {2020},
pages = {3610-3620},
volume = {108},
url = {https://mlanthology.org/aistats/2020/mialon2020aistats-screening/}
}