More Powerful Selective Kernel Tests for Feature Selection

TL;DR

This paper introduces enhanced selective kernel tests for feature selection that condition on minimal selection events, improving statistical power while maintaining control over false positive rates, demonstrated through synthetic and real data experiments.

Contribution

It extends recent feature selection methods by incorporating minimal conditioning events using multiscale bootstrap, leading to more powerful tests.

Findings

Proposed tests outperform existing methods in power across various scenarios.

The approach maintains controlled false positive rates.

Experimental validation on synthetic and real datasets confirms effectiveness.

Abstract

Refining one's hypotheses in the light of data is a common scientific practice; however, the dependency on the data introduces selection bias and can lead to specious statistical analysis. An approach for addressing this is via conditioning on the selection procedure to account for how we have used the data to generate our hypotheses, and prevent information to be used again after selection. Many selective inference (a.k.a. post-selection inference) algorithms typically take this approach but will "over-condition" for sake of tractability. While this practice yields well calibrated statistic tests with controlled false positive rates (FPR), it can incur a major loss in power. In our work, we extend two recent proposals for selecting features using the Maximum Mean Discrepancy and Hilbert Schmidt Independence Criterion to condition on the minimal conditioning event. We show how recent advances in multiscale bootstrap makes conditioning on the minimal selection event possible and demonstrate our proposal over a range of synthetic and real world experiments. Our results show that our proposed test is indeed more powerful in most scenarios.