FDGATII : Fast Dynamic Graph Attention with Initial Residual and Identity Mapping

TL;DR

FDGATII is a novel graph neural network that effectively addresses over-smoothing, noisy neighbors, and the suspended animation problem using dynamic attention, residuals, and identity mapping, achieving state-of-the-art results on multiple datasets.

Contribution

Introduces FDGATII, a scalable, efficient GNN combining dynamic self-attention with residuals and identity mapping to improve performance on heterophilic graphs.

Findings

Outperforms GAT and GCN benchmarks in accuracy.

Achieves state-of-the-art results on Chameleon and Cornell datasets.

Demonstrates scalability and versatility across multiple datasets.

Abstract

While Graph Neural Networks have gained popularity in multiple domains, graph-structured input remains a major challenge due to (a) over-smoothing, (b) noisy neighbours (heterophily), and (c) the suspended animation problem. To address all these problems simultaneously, we propose a novel graph neural network FDGATII, inspired by attention mechanism's ability to focus on selective information supplemented with two feature preserving mechanisms. FDGATII combines Initial Residuals and Identity Mapping with the more expressive dynamic self-attention to handle noise prevalent from the neighbourhoods in heterophilic data sets. By using sparse dynamic attention, FDGATII is inherently parallelizable in design, whist efficient in operation; thus theoretically able to scale to arbitrary graphs with ease. Our approach has been extensively evaluated on 7 datasets. We show that FDGATII outperforms GAT and GCN based benchmarks in accuracy and performance on fully supervised tasks, obtaining state-of-the-art results on Chameleon and Cornell datasets with zero domain-specific graph pre-processing, and demonstrate its versatility and fairness.