Learning Conjoint Attentions for Graph Neural Nets
This work provides an incremental improvement in GNN performance for researchers and practitioners working on graph-structured data by incorporating structural interventions into the attention mechanism.
This paper introduces Conjoint Attentions (CAs), a new attention mechanism for Graph Neural Networks (GNNs) that integrates structural information beyond node features. The proposed Graph Conjoint Attention Networks (CATs) leverage CAs to learn representations, demonstrating notable performance improvements on benchmarking datasets compared to state-of-the-art baselines.
In this paper, we present Conjoint Attentions (CAs), a class of novel learning-to-attend strategies for graph neural networks (GNNs). Besides considering the layer-wise node features propagated within the GNN, CAs can additionally incorporate various structural interventions, such as node cluster embedding, and higher-order structural correlations that can be learned outside of GNN, when computing attention scores. The node features that are regarded as significant by the conjoint criteria are therefore more likely to be propagated in the GNN. Given the novel Conjoint Attention strategies, we then propose Graph conjoint attention networks (CATs) that can learn representations embedded with significant latent features deemed by the Conjoint Attentions. Besides, we theoretically validate the discriminative capacity of CATs. CATs utilizing the proposed Conjoint Attention strategies have been extensively tested in well-established benchmarking datasets and comprehensively compared with state-of-the-art baselines. The obtained notable performance demonstrates the effectiveness of the proposed Conjoint Attentions.