Consistent feature selection for neural networks via Adaptive Group Lasso

TL;DR

This paper introduces a theoretically grounded method using adaptive group lasso for feature selection in neural networks, enhancing interpretability by reliably identifying significant features in single-hidden-layer models.

Contribution

It provides the first theoretical guarantee for adaptive group lasso in neural network feature selection, demonstrating consistency in single-hidden-layer models.

Findings

Method is consistent for feature selection in neural networks.

Validated through simulation and real data analysis.

Enhances interpretability of neural network models.

Abstract

One main obstacle for the wide use of deep learning in medical and engineering sciences is its interpretability. While neural network models are strong tools for making predictions, they often provide little information about which features play significant roles in influencing the prediction accuracy. To overcome this issue, many regularization procedures for learning with neural networks have been proposed for dropping non-significant features. Unfortunately, the lack of theoretical results casts doubt on the applicability of such pipelines. In this work, we propose and establish a theoretical guarantee for the use of the adaptive group lasso for selecting important features of neural networks. Specifically, we show that our feature selection method is consistent for single-output feed-forward neural networks with one hidden layer and hyperbolic tangent activation function. We demonstrate its applicability using both simulation and data analysis.