Timezone: »
Graph contrastive learning (GCL) is a major self-supervised graph learning technique that aims to capture invariant properties of graphs via instance discrimination. Its performance heavily relies on the construction of multiple graph views yet it still remains unclear about what makes effective topology augmentations. Recent studies mainly perform topology augmentations in a uniformly random manner without considering graph properties. In this work, we aim to find principled ways for topology augmentations by exploring the invariance of graphs from the graph spectral perspective. Specifically, we propose a novel topology augmentation method guided by spectral change. Extensive experiments on both graph and node classification tasks demonstrate the effectiveness of our method in capturing the structural essence of graphs for self-supervised learning. The proposed method also brings promising performance in transfer learning and adversarial attack settings. We envision this work to provide a principled way for graph augmentation.
Author Information
Lu Lin (Penn State University)
Jinghui Chen (Penn State University)
Hongning Wang (University of Virginia)
More from the Same Authors
-
2022 : On the Vulnerability of Backdoor Defenses for Federated Learning »
Pei Fang · Jinghui Chen -
2022 : Accelerating Adaptive Federated Optimization with Local Gossip Communications »
Yujia Wang · Pei Fang · Jinghui Chen -
2022 : How Powerful is Implicit Denoising in Graph Neural Networks »
Songtao Liu · Rex Ying · Hanze Dong · Lu Lin · Jinghui Chen · Dinghao Wu -
2022 Poster: Communication Efficient Distributed Learning for Kernelized Contextual Bandits »
Chuanhao Li · Huazheng Wang · Mengdi Wang · Hongning Wang -
2022 Poster: One-shot Neural Backdoor Erasing via Adversarial Weight Masking »
Shuwen Chai · Jinghui Chen -
2022 Poster: Communication Efficient Federated Learning for Generalized Linear Bandits »
Chuanhao Li · Hongning Wang -
2019 Poster: Model-Based Reinforcement Learning with Adversarial Training for Online Recommendation »
Xueying Bai · Jian Guan · Hongning Wang -
2018 Poster: Bandit Learning with Implicit Feedback »
Yi Qi · Qingyun Wu · Hongning Wang · Jie Tang · Maosong Sun