Sharpness-Aware Graph Collaborative Filtering

TL;DR

This paper introduces gSAM, a training method for Graph Neural Networks in collaborative filtering that encourages convergence to flatter minima, leading to better generalization and improved performance.

Contribution

The paper proposes gSAM, a novel bi-level optimization approach that regularizes flatness in GNN training to enhance generalization in collaborative filtering.

Findings

gSAM outperforms standard training methods in experiments.

Flatter minima correlate with better test performance.

gSAM effectively escapes sharp minima during training.

Abstract

Graph Neural Networks (GNNs) have achieved impressive performance in collaborative filtering. However, GNNs tend to yield inferior performance when the distributions of training and test data are not aligned well. Also, training GNNs requires optimizing non-convex neural networks with an abundance of local and global minima, which may differ widely in their performance at test time. Thus, it is essential to choose the minima carefully. Here we propose an effective training schema, called {gSAM}, under the principle that the \textit{flatter} minima has a better generalization ability than the \textit{sharper} ones. To achieve this goal, gSAM regularizes the flatness of the weight loss landscape by forming a bi-level optimization: the outer problem conducts the standard model training while the inner problem helps the model jump out of the sharp minima. Experimental results show the superiority of our gSAM.