Adaptive Autoguidance for Item-Side Fairness in Diffusion Recommender Systems

2026-02-16

AI Summary

Diffusion-based recommender systems are prone to popularity bias, which often leads to insufficient exposure of long-tail items. To address this issue, this work proposes A2G-DiffRec, a novel framework that enhances item-side fairness by dynamically fusing the outputs of the main diffusion model and its under-trained counterpart through an adaptive auto-guidance mechanism. Unlike conventional approaches that rely on fixed guidance weights, A2G-DiffRec employs an adaptive weighting strategy, enabling a more effective trade-off between recommendation accuracy and fairness. Additionally, the method incorporates a popularity-aware regularization constraint to further mitigate exposure disparity. Experimental results on three real-world datasets—MovieLens-1M, Foursquare-Tokyo, and Music4All-Onion—demonstrate that A2G-DiffRec significantly improves item exposure fairness while incurring only a marginal drop in recommendation accuracy.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Data Mining & Knowledge Management: Recommender Systems
• Machine Learning: Ethics, Bias, and Fairness

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Web Mining and Content Analysis: Content-based information diffusion
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Diffusion recommender systems achieve strong recommendation accuracy but often suffer from popularity bias, resulting in unequal item exposure. To address this shortcoming, we introduce A2G-DiffRec, a diffusion recommender that incorporates adaptive autoguidance, where the main model is guided by a less-trained version of itself. Instead of using a fixed guidance weight, A2G-DiffRec learns to adaptively weigh the outputs of the main and weak models during training, supervised by a popularity regularization that promotes balanced exposure across items with different popularity levels. Experimental results on the MovieLens-1M, Foursquare-Tokyo, and Music4All-Onion datasets show that A2G-DiffRec is effective in enhancing item-side fairness at a marginal cost of accuracy reduction compared to existing guided diffusion recommenders and other non-diffusion baselines.

Problem

• popularity bias
• item-side fairness
• diffusion recommender systems
• unequal item exposure

Innovation

• adaptive autoguidance
• diffusion recommender systems
• item-side fairness
• popularity bias
• popularity regularization