Discrepancy-Aware Graph Mask Auto-Encoder
This addresses a limitation in graph self-supervised learning for heterophilic graphs, offering a novel approach with broad applicability in graph analytics.
The paper tackles the problem of masked graph auto-encoders failing to generalize to heterophilic graphs by proposing DGMAE, which reconstructs discrepancy information to obtain more distinguishable node representations, significantly outperforming state-of-the-art methods on 17 benchmark datasets across three graph analytic tasks.
Masked Graph Auto-Encoder, a powerful graph self-supervised training paradigm, has recently shown superior performance in graph representation learning. Existing works typically rely on node contextual information to recover the masked information. However, they fail to generalize well to heterophilic graphs where connected nodes may be not similar, because they focus only on capturing the neighborhood information and ignoring the discrepancy information between different nodes, resulting in indistinguishable node representations. In this paper, to address this issue, we propose a Discrepancy-Aware Graph Mask Auto-Encoder (DGMAE). It obtains more distinguishable node representations by reconstructing the discrepancy information of neighboring nodes during the masking process. We conduct extensive experiments on 17 widely-used benchmark datasets. The results show that our DGMAE can effectively preserve the discrepancies of nodes in low-dimensional space. Moreover, DGMAE significantly outperforms state-of-the-art graph self-supervised learning methods on three graph analytic including tasks node classification, node clustering, and graph classification, demonstrating its remarkable superiority. The code of DGMAE is available at https://github.com/zhengziyu77/DGMAE.