Beyond Graph Convolution: Multimodal Recommendation with Topology-aware MLPs

TL;DR

This paper introduces TMLP, a topology-aware MLP model that effectively models multimodal item relations without GCNs, improving recommendation accuracy, efficiency, and robustness in multimodal recommender systems.

Contribution

The paper proposes TMLP, a novel GCN-free approach that leverages topological pruning and intra/inter-modality learning to better capture complex multimodal item relationships.

Findings

TMLP outperforms nine baseline models on three real-world datasets.

TMLP achieves higher training efficiency and robustness compared to GCN-based models.

Discarding GCN message passing improves model performance and stability.

Abstract

Given the large volume of side information from different modalities, multimodal recommender systems have become increasingly vital, as they exploit richer semantic information beyond user-item interactions. Recent works highlight that leveraging Graph Convolutional Networks (GCNs) to explicitly model multimodal item-item relations can significantly enhance recommendation performance. However, due to the inherent over-smoothing issue of GCNs, existing models benefit only from shallow GCNs with limited representation power. This drawback is especially pronounced when facing complex and high-dimensional patterns such as multimodal data, as it requires large-capacity models to accommodate complicated correlations. To this end, in this paper, we investigate bypassing GCNs when modeling multimodal item-item relationship. More specifically, we propose a Topology-aware Multi-Layer Perceptron (TMLP), which uses MLPs instead of GCNs to model the relationships between items. TMLP enhances MLPs with topological pruning to denoise item-item relations and intra (inter)-modality learning to integrate higher-order modality correlations. Extensive experiments on three real-world datasets verify TMLP's superiority over nine baselines. We also find that by discarding the internal message passing in GCNs, which is sensitive to node connections, TMLP achieves significant improvements in both training efficiency and robustness against existing models.