GraphNAS: Graph Neural Architecture Search with Reinforcement Learning

TL;DR

GraphNAS automates the design of graph neural networks using reinforcement learning, significantly improving performance on node classification tasks across various datasets.

Contribution

This paper introduces a reinforcement learning-based method for automatic graph neural architecture search, reducing manual effort and achieving competitive results.

Findings

GraphNAS outperforms existing manually designed GNNs on multiple datasets.

The method can discover architectures comparable to human-designed models.

Experimental results show consistent performance improvements.

Abstract

Graph Neural Networks (GNNs) have been popularly used for analyzing non-Euclidean data such as social network data and biological data. Despite their success, the design of graph neural networks requires a lot of manual work and domain knowledge. In this paper, we propose a Graph Neural Architecture Search method (GraphNAS for short) that enables automatic search of the best graph neural architecture based on reinforcement learning. Specifically, GraphNAS first uses a recurrent network to generate variable-length strings that describe the architectures of graph neural networks, and then trains the recurrent network with reinforcement learning to maximize the expected accuracy of the generated architectures on a validation data set. Extensive experimental results on node classification tasks in both transductive and inductive learning settings demonstrate that GraphNAS can achieve consistently better performance on the Cora, Citeseer, Pubmed citation network, and protein-protein interaction network. On node classification tasks, GraphNAS can design a novel network architecture that rivals the best human-invented architecture in terms of test set accuracy.