AISTATS 2020poster6 citations
Differentiable Feature Selection by Discrete Relaxation
Abstract
In this paper, we introduce Differentiable Feature Selection, a gradient-based search algorithm for feature selection. Our approach extends a recent result on the estimation of learnability in the sublinear data regime by showing that the calculation can be performed iteratively (i.e. in mini-batches) and in linear time and space with respect to both the number of features D and the sample size N. This, along with a discrete-to-continuous relaxation of the search domain, allows for an efficient, gradient-based search algorithm among feature subsets for very large datasets. Our algorithm utilizes higher-order correlations between features and targets for both the N>D and N
BibTeX
@InProceedings{pmlr-v108-sheth20a,
title = {Differentiable Feature Selection by Discrete Relaxation},
author = {Sheth, Rishit and Fusi, Nicol\'o},
booktitle = {Proceedings of the Twenty Third International Conference on Artificial Intelligence and Statistics},
pages = {1564--1572},
year = {2020},
editor = {Chiappa, Silvia and Calandra, Roberto},
volume = {108},
series = {Proceedings of Machine Learning Research},
month = {26--28 Aug},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v108/sheth20a/sheth20a.pdf},
url = {https://proceedings.mlr.press/v108/sheth20a.html},
abstract = {In this paper, we introduce Differentiable Feature Selection, a gradient-based search algorithm for feature selection. Our approach extends a recent result on the estimation of learnability in the sublinear data regime by showing that the calculation can be performed iteratively (i.e. in mini-batches) and in linear time and space with respect to both the number of features D and the sample size N. This, along with a discrete-to-continuous relaxation of the search domain, allows for an efficient, gradient-based search algorithm among feature subsets for very large datasets. Our algorithm utilizes higher-order correlations between features and targets for both the N>D and N