Learning to Maximize Mutual Information for Dynamic Feature Selection

Author name not available (Why is that?)

Publication date: 2 January 2023

Abstract: Feature selection helps reduce data acquisition costs in ML, but the standard approach is to train models with static feature subsets. Here, we consider the dynamic feature selection (DFS) problem where a model sequentially queries features based on the presently available information. DFS is often addressed with reinforcement learning, but we explore a simpler approach of greedily selecting features based on their conditional mutual information. This method is theoretically appealing but requires oracle access to the data distribution, so we develop a learning approach based on amortized optimization. The proposed method is shown to recover the greedy policy when trained to optimality, and it outperforms numerous existing feature selection methods in our experiments, thus validating it as a simple but powerful approach for this problem.

Has companion code repository: https://github.com/iancovert/dynamic-selection

This page was built for publication: Learning to Maximize Mutual Information for Dynamic Feature Selection

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6422300)