Towards Deep Attention in Graph Neural Networks: Problems and Remedies

TL;DR

This paper investigates the challenges of deep graph attention in GNNs, identifies key problems like over-smoothing, and proposes AERO-GNN, a new architecture that effectively mitigates these issues and improves performance on node classification tasks.

Contribution

The paper introduces AERO-GNN, a novel deep graph attention architecture that addresses over-smoothing and attention smoothness issues, with theoretical guarantees and empirical validation.

Findings

AERO-GNN mitigates over-smoothing in deep layers.

It achieves higher accuracy on multiple node classification benchmarks.

Deep attention in GNNs can be effectively stabilized with the proposed methods.

Abstract

Graph neural networks (GNNs) learn the representation of graph-structured data, and their expressiveness can be further enhanced by inferring node relations for propagation. Attention-based GNNs infer neighbor importance to manipulate the weight of its propagation. Despite their popularity, the discussion on deep graph attention and its unique challenges has been limited. In this work, we investigate some problematic phenomena related to deep graph attention, including vulnerability to over-smoothed features and smooth cumulative attention. Through theoretical and empirical analyses, we show that various attention-based GNNs suffer from these problems. Motivated by our findings, we propose AEROGNN, a novel GNN architecture designed for deep graph attention. AERO-GNN provably mitigates the proposed problems of deep graph attention, which is further empirically demonstrated with (a) its adaptive and less smooth attention functions and (b) higher performance at deep layers (up to 64). On 9 out of 12 node classification benchmarks, AERO-GNN outperforms the baseline GNNs, highlighting the advantages of deep graph attention. Our code is available at https://github.com/syleeheal/AERO-GNN.