An Automatic Relevance Determination Prior Bayesian Neural Network for Controlled Variable Selection

This work addresses variable selection challenges in fields like environmental science, offering incremental improvements over existing methods.

The authors tackled the problem of variable selection in high-dimensional data by proposing a Bayesian Neural Network with Automatic Relevance Determination prior as a feature importance statistic for the model-x knockoff filter, resulting in statistically significant improvements in variable selection power and predictive performance on simulated and real-world datasets.