FairFS: Addressing Deep Feature Selection Biases for Recommender System

2026-02-23

AI Summary

This work addresses the challenge of inaccurate feature importance estimation in deep recommender systems, which is often compromised by layer bias, baseline bias, and approximation bias. To mitigate these issues collectively, the authors propose FairFS, a novel algorithm that systematically identifies and alleviates all three biases: it employs multi-layer regularization to harmonize feature contributions across nonlinear transformations, designs a smooth baseline proximal to the decision boundary, and introduces a gradient aggregation strategy for more faithful approximation. Integrated with a trainable gating mechanism and sensitivity analysis, FairFS significantly enhances both the fairness and accuracy of feature importance evaluation. Extensive experiments on multiple real-world datasets demonstrate that FairFS achieves state-of-the-art performance in feature selection.

Technology Category

• Machine Learning: Dimensionality Reduction/Feature Selection
• Computer Vision: Bias, Fairness & Privacy
• Data Mining & Knowledge Management: Recommender Systems

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Social Networks and Social Media: Fairness and bias in social network and social media analysis

Abstract

Large-scale online marketplaces and recommender systems serve as critical technological support for e-commerce development. In industrial recommender systems, features play vital roles as they carry information for downstream models. Accurate feature importance estimation is critical because it helps identify the most useful feature subsets from thousands of feature candidates for online services. Such selection enables improved online performance while reducing computational cost. To address feature selection problems in deep learning, trainable gate-based and sensitivity-based methods have been proposed and proven effective in industrial practice. However, through the analysis of real-world cases, we identified three bias issues that cause feature importance estimation to rely on partial model layers, samples, or gradients, ultimately leading to inaccurate importance estimation. We refer to these as layer bias, baseline bias, and approximation bias. To mitigate these issues, we propose FairFS, a fair and accurate feature selection algorithm. FairFS regularizes feature importance estimated across all nonlinear transformation layers to address layer bias. It also introduces a smooth baseline feature close to the classifier decision boundary and adopts an aggregated approximation method to alleviate baseline and approximation biases. Extensive experiments demonstrate that FairFS effectively mitigates these biases and achieves state-of-the-art feature selection performance.

Problem

• feature selection bias
• layer bias
• baseline bias
• approximation bias
• feature importance estimation

Innovation

• FairFS
• feature selection bias
• layer bias
• baseline bias
• approximation bias