Automated Self-Supervised Learning for Graphs

TL;DR

This paper introduces AutoSSL, a framework that automatically searches for optimal combinations of self-supervised pretext tasks for graph learning, leveraging homophily to improve downstream node classification and clustering performance.

Contribution

It proposes a novel automatic search method for combining multiple self-supervised tasks in graph learning, guided by the principle of homophily, with theoretical and empirical validation.

Findings

AutoSSL outperforms single-task methods on 7 datasets.

Significant improvements in node classification and clustering accuracy.

The framework effectively leverages multiple pretext tasks for better representations.

Abstract

Graph self-supervised learning has gained increasing attention due to its capacity to learn expressive node representations. Many pretext tasks, or loss functions have been designed from distinct perspectives. However, we observe that different pretext tasks affect downstream tasks differently cross datasets, which suggests that searching pretext tasks is crucial for graph self-supervised learning. Different from existing works focusing on designing single pretext tasks, this work aims to investigate how to automatically leverage multiple pretext tasks effectively. Nevertheless, evaluating representations derived from multiple pretext tasks without direct access to ground truth labels makes this problem challenging. To address this obstacle, we make use of a key principle of many real-world graphs, i.e., homophily, or the principle that "like attracts like," as the guidance to effectively search various self-supervised pretext tasks. We provide theoretical understanding and empirical evidence to justify the flexibility of homophily in this search task. Then we propose the AutoSSL framework which can automatically search over combinations of various self-supervised tasks. By evaluating the framework on 7 real-world datasets, our experimental results show that AutoSSL can significantly boost the performance on downstream tasks including node clustering and node classification compared with training under individual tasks. Code is released at https://github.com/ChandlerBang/AutoSSL.