We are pleased to announce that a groundbreaking conference paper titled “Simple yet Effective Gradient-Free Graph Convolutional Networks” will be presented at the International Joint Conference on Neural Networks in September 2025. This work is led by Dr. Yulin Zhu from our institute as the first author, in collaboration with The Hong Kong Polytechnic University.
- Yulin Zhu, Xing Ai, Qimai Li, Xiaoming Wu, Wai-Lun Lo, Kai Zhou. (2025, June 30th-July 5th ). Simple yet Effective Gradient-Free Graph Convolutional Networks. International Joint Conference on Neural Networks, Rome, Italy.
Paper Abstract
Linearized Graph Neural Networks (GNNs) have attracted great attention in recent years for graph representation learning. Compared with nonlinear Graph Neural Network (GNN) models, linearized GNNs are much more time-efficient and can achieve comparable performances on typical downstream tasks such as node classification. Although some linearized GNN variants are purposely crafted to mitigate “over-smoothing”, empirical studies demonstrate that they still somehow suffer from this issue. In this paper, we instead relate over-smoothing with the vanishing gradient phenomenon and craft a gradient-free training framework to achieve more efficient and effective linearized GNNs which can significantly overcome over-smoothing and enhance the generalization of the model. The experimental results demonstrate that our methods achieve better and more stable performances on node classification tasks with varying depths and cost much less training time.
The research team members
Hong Kong Chu Hai College
- Dr. Yulin Zhu — Assistant Professor, Department of Computer Science
- Prof. Wai Lun Lo — Full Professor, Director of Quality Assurance, Head of Department of Computer Science
The Hong Kong Polytechnic University
- Xing Ai — PhD student
- Qimai Li — PhD Graduate，Quantitative trading engineer at JIWEI Fund
- Xiaoming Wu — Associate Professor, Department of Data Science and Artificial Intelligence
- Kai Zhou — Assistant Professor, Department of Computing Science
Photos from the paper
