Subgraph Sampling
Subgraph sampling techniques aim to efficiently process large graphs by focusing on smaller, representative subgraphs for tasks like node classification and link prediction within graph neural networks (GNNs). Current research emphasizes developing sophisticated sampling strategies that account for graph characteristics like homophily and heterophily, and optimizing these strategies for both speed and accuracy, often incorporating techniques like submodularity and contrastive learning. These advancements are crucial for scaling GNNs to massive datasets and improving their performance in various applications, including molecular property prediction and decentralized learning over wireless networks.
Papers
Self-Supervised Dynamic Graph Representation Learning via Temporal Subgraph Contrast
Linpu Jiang, Ke-Jia Chen, Jingqiang Chen
BGL: GPU-Efficient GNN Training by Optimizing Graph Data I/O and Preprocessing
Tianfeng Liu, Yangrui Chen, Dan Li, Chuan Wu, Yibo Zhu, Jun He, Yanghua Peng, Hongzheng Chen, Hongzhi Chen, Chuanxiong Guo