DrGNN: Deep Residual Graph Neural Network with Contrastive Learning

Recent studies reveal that deep representation learning models without proper regularization can suffer from the dimensional col lapse problem, i.e., representation vectors span over a lower dimensional space. In the domain of graph deep representation learning, the phenomenon that the node representations are indistinguishable and even shrink to a constant vector is called oversmoothing. Based on the analysis of the rank of node representations, we find that representation oversmoothing and dimensional collapse are highly related to each other in deep graph neural networks, and the oversmoothing problem can be interpreted by the dimensional collapse of the node representation matrix. Then, to address the dimensional collapse and the oversmoothing together in deep graph neural networks, we first find vanilla residual connections and contrastive learning producing sub-optimal outcomes by ignoring the structured constraints of graph data. Motivated by this, we propose a novel graph neural network named DrGNN to alleviate the oversmoothing issue from the perspective of addressing dimensional collapse. Specifically, in DrGNN, we design a topology-preserving residual connection for graph neural networks to force the low-rank of hidden representations close to the full-rank input features. Also, we propose the structure-guided contrastive learning to ensure only close neighbors who share similar local connections can have similar representations. Empirical experiments on multiple real-world datasets demonstrate that DrGNN outperforms state-of-the-art deep graph representation baseline algorithms. The code of our method is available at the GitHub link: https://github.com/zhenglecheng/DrGNN.