ASPIRE: Make Spectral Graph Collaborative Filtering Great Again via Adaptive Filter Learning
Yunhang He, Cong Xu, Zhangchi Zhu, Hongzhi Yin, Wei Zhang
TLDR
ASPIRE introduces an adaptive filter learning framework for spectral graph collaborative filtering, overcoming manual tuning and "low-frequency explosion."
Key contributions
- Identifies "low-frequency explosion" bias in traditional recommendation objectives hindering effective filter learning.
- Proposes ASPIRE, an adaptive spectral graph collaborative filtering framework using bi-level optimization.
- Disentangles the filter learning objective based on theoretical analysis for better performance and stability.
- Achieves excellent recommendation performance, matching engineered designs and effective in LLM-powered CF.
Why it matters
This paper makes spectral graph collaborative filtering more effective by introducing fully learnable graph filters, overcoming a critical "low-frequency explosion" bias. ASPIRE's adaptive framework improves recommendation performance and stability, demonstrating the viability of generalizable filter learning for expressive GNNs.
Original Abstract
Graph filter design is central to spectral collaborative filtering, yet most existing methods rely on manually tuned hyperparameters rather than fully learnable filters. We show that this challenge stems from a bias in traditional recommendation objectives, which induces a spectral phenomenon termed low-frequency explosion, thereby fundamentally hindering the effective learning of graph filters. To overcome this limitation, we propose a novel adaptive spectral graph collaborative filtering framework (ASPIRE) based on a bi-level optimization objective. Guided by our theoretical analysis, we disentangle the filter learning objective, which in turn leads to excellent recommendation performance, spectral adaptivity, and training stability in practice. Extensive experiments show our learned filters match the performance of carefully engineered task-specific designs. Furthermore, ASPIRE is equally effective in LLM-powered collaborative filtering. Our findings demonstrate that graph filter learning is viable and generalizable, paving the way for more expressive graph neural networks in collaborative filtering.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.