Hyperbolic Diffusion Recommender Model
Diffusion models (DMs) have emerged as the new state-of-the-art family of deep generative models. To gain deeper insights into the limitations of diffusion models in recommender systems, we investigate the fundamental structural disparities between images and items. Consequently, items often exhibit distinct anisotropic and directional structures that are less prevalent in images. However, the traditional forward diffusion process continuously adds isotropic Gaussian noise, causing anisotropic signals to degrade into noise, which impairs the semantically meaningful representations in recommender systems.
Authors
- Deqing Wang (ORCID: https://orcid.org/0000-0003-3463-9702)
- Wei Chen (ORCID: https://orcid.org/0000-0001-7454-3882)
- Fuzhen Zhuang (ORCID: https://orcid.org/0000-0001-9170-7009)
- Meng Yuan (ORCID: https://orcid.org/0000-0003-4016-5048)
- Yutian Xiao (ORCID: https://orcid.org/0000-0002-8276-7920)
- Chu Zhao (ORCID: https://orcid.org/0000-0002-5649-1545)
Institutions
- Chongqing University of Technology (CN)
- Beihang University (CN)
Publication Details
- Published
- 2025-04-22
- DOI
- https://doi.org/10.1145/3696410.3714873
- Citations
- 5
- Primary Topic
- Recommender Systems and Techniques
- Type
- article
- Field-Weighted Citation Impact
- 16.23
Funders
- National Natural Science Foundation of China
- National Key Research and Development Program of China
- Fundamental Research Funds for the Central Universities