ASPIRE: Make Spectral Graph Collaborative Filtering Great Again via Adaptive Filter Learning

TL;DR

ASPIRE introduces an adaptive spectral graph filtering framework that learns filters directly, overcoming traditional biases and enhancing collaborative filtering performance across various settings.

Contribution

It proposes a bi-level optimization approach for fully learnable graph filters, addressing spectral biases and improving recommendation accuracy and stability.

Findings

Learned filters match engineered designs in performance.

ASPIRE improves spectral adaptivity and training stability.

Effective in both traditional and LLM-powered collaborative filtering.

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.