AMOSL: Adaptive Modality-wise Structure Learning in Multi-view Graph Neural Networks For Enhanced Unified Representation

TL;DR

AMoSL introduces an adaptive structure learning method for multi-view graph neural networks, addressing modality discrepancies and improving classification accuracy through optimal transport and joint embedding.

Contribution

It proposes a novel adaptive modality-wise structure learning approach that captures node correspondences and enhances multi-view GNNs with efficient bilevel optimization.

Findings

Improves graph classification accuracy on six benchmark datasets.

Effectively captures inter-modality node correspondences.

Enhances robustness of multi-view GNNs to modality discrepancies.

Abstract

While Multi-view Graph Neural Networks (MVGNNs) excel at leveraging diverse modalities for learning object representation, existing methods assume identical local topology structures across modalities that overlook real-world discrepancies. This leads MVGNNs straggles in modality fusion and representations denoising. To address these issues, we propose adaptive modality-wise structure learning (AMoSL). AMoSL captures node correspondences between modalities via optimal transport, and jointly learning with graph embedding. To enable efficient end-to-end training, we employ an efficient solution for the resulting complex bilevel optimization problem. Furthermore, AMoSL adapts to downstream tasks through unsupervised learning on inter-modality distances. The effectiveness of AMoSL is demonstrated by its ability to train more accurate graph classifiers on six benchmark datasets.