Graph neural networks (GNNs) emerged recently as a powerful tool for analyzing non-Euclidean data such as social network data. Despite their success, the design of graph neural networks requires heavy manual work and domain knowledge. In this paper, we present a graph neural architecture search method (GraphNAS) that enables automatic design of the best graph neural architecture based on reinforcement learning. Specifically, GraphNAS uses a recurrent network to generate variable-length strings that describe the architectures of graph neural networks, and trains the recurrent network with policy gradient to maximize the expected accuracy of the generated architectures on a validation data set. Furthermore, to improve the search efficiency of GraphNAS on big networks, GraphNAS restricts the search space from an entire architecture space to a sequential concatenation of the best search results built on each single architecture layer. Experiments on real-world datasets demonstrate that GraphNAS can design a novel network architecture that rivals the best human-invented architecture in terms of validation set accuracy. Moreover, in a transfer learning task we observe that graph neural architectures designed by GraphNAS, when transferred to new datasets, still gain improvement in terms of prediction accuracy.
CITATION STYLE
Gao, Y., Yang, H., Zhang, P., Zhou, C., & Hu, Y. (2020). Graph neural architecture search. In IJCAI International Joint Conference on Artificial Intelligence (Vol. 2021-January, pp. 1403–1409). International Joint Conferences on Artificial Intelligence. https://doi.org/10.24963/ijcai.2020/195
Mendeley helps you to discover research relevant for your work.