Asymmetric Diffusion Recommendation Model

2025-08-18

AI Summary

Mainstream diffusion models assume continuous data spaces and rely on symmetric Gaussian noise, which is ill-suited for inherently discrete user behavior data in recommender systems and leads to loss of personalized information. Method: We propose the first asymmetric diffusion model tailored for discrete recommendation scenarios. It introduces a generalized forward process to simulate realistic feature missing patterns, constructs a reverse denoising mechanism in an asymmetric latent space, and incorporates a task-aware optimization strategy that explicitly preserves personalized representations during generative learning. Contribution/Results: Extensive offline experiments validate its effectiveness. Deployed on the Douyin Music App, online A/B testing shows statistically significant improvements: +0.131% in active user days and +0.166% in average session duration. This work pioneers the integration of asymmetric diffusion mechanisms into recommender systems, establishing a novel paradigm for robust generative representation learning on discrete data.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Data Mining & Knowledge Management: Recommender Systems

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• 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

Recently, motivated by the outstanding achievements of diffusion models, the diffusion process has been employed to strengthen representation learning in recommendation systems. Most diffusion-based recommendation models typically utilize standard Gaussian noise in symmetric forward and reverse processes in continuous data space. Nevertheless, the samples derived from recommendation systems inhabit a discrete data space, which is fundamentally different from the continuous one. Moreover, Gaussian noise has the potential to corrupt personalized information within latent representations. In this work, we propose a novel and effective method, named Asymmetric Diffusion Recommendation Model (AsymDiffRec), which learns forward and reverse processes in an asymmetric manner. We define a generalized forward process that simulates the missing features in real-world recommendation samples. The reverse process is then performed in an asymmetric latent feature space. To preserve personalized information within the latent representation, a task-oriented optimization strategy is introduced. In the serving stage, the raw sample with missing features is regarded as a noisy input to generate a denoising and robust representation for the final prediction. By equipping base models with AsymDiffRec, we conduct online A/B tests, achieving improvements of +0.131% and +0.166% in terms of users' active days and app usage duration respectively. Additionally, the extended offline experiments also demonstrate improvements. AsymDiffRec has been implemented in the Douyin Music App.

Problem

• Handles discrete data space in recommendation systems
• Preserves personalized information in latent representations
• Improves recommendation accuracy with asymmetric diffusion processes

Innovation

• Asymmetric forward and reverse diffusion processes
• Generalized forward process for missing features
• Task-oriented optimization preserves personalized information