Normalizing Flows for Knockoff-free Controlled Feature Selection
This addresses the problem of reliable feature selection with statistical guarantees for researchers and practitioners in fields like genetics, where false discoveries can mislead conclusions.
The paper tackles the problem of controlled feature selection with false discovery rate (FDR) control, showing that existing deep-learning-based knockoff methods often fail to control FDR due to inaccurate feature modeling and poor enforcement of the swap property. The result is FlowSelect, which uses normalizing flows and a novel MCMC-based procedure to compute valid p-values, empirically controlling FDR on benchmarks and demonstrating greater power, including in genetic variant inference from GWAS data.
Controlled feature selection aims to discover the features a response depends on while limiting the false discovery rate (FDR) to a predefined level. Recently, multiple deep-learning-based methods have been proposed to perform controlled feature selection through the Model-X knockoff framework. We demonstrate, however, that these methods often fail to control the FDR for two reasons. First, these methods often learn inaccurate models of features. Second, the "swap" property, which is required for knockoffs to be valid, is often not well enforced. We propose a new procedure called FlowSelect to perform controlled feature selection that does not suffer from either of these two problems. To more accurately model the features, FlowSelect uses normalizing flows, the state-of-the-art method for density estimation. Instead of enforcing the "swap" property, FlowSelect uses a novel MCMC-based procedure to calculate p-values for each feature directly. Asymptotically, FlowSelect computes valid p-values. Empirically, FlowSelect consistently controls the FDR on both synthetic and semi-synthetic benchmarks, whereas competing knockoff-based approaches do not. FlowSelect also demonstrates greater power on these benchmarks. Additionally, FlowSelect correctly infers the genetic variants associated with specific soybean traits from GWAS data.