The content introduces the Graph Signal Diffusion Model for Collaborative Filtering, highlighting the limitations of existing diffusion models in handling implicit feedback data. The proposed GiffCF model utilizes graph signal processing techniques to smooth and sharpen interaction signals, improving recommendation performance. The forward process involves heat equation-based graph smoothing, while the reverse process iteratively refines and sharpens preference signals. Extensive experiments demonstrate the effectiveness of GiffCF on benchmark datasets.
Naar een andere taal
vanuit de broninhoud
arxiv.org
Belangrijkste Inzichten Gedestilleerd Uit
by Yunqin Zhu,C... om arxiv.org 03-27-2024
https://arxiv.org/pdf/2311.08744.pdfDiepere vragen