Generating Counterfactual Hard Negative Samples for Graph Contrastive Learning
Open Access
- 30 April 2023
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 621-629
Abstract
Graph contrastive learning has emerged as a powerful unsupervised graph representation learning tool. The key to the success of graph contrastive learning is to acquire high-quality positive and negative samples as contrasting pairs to learn the underlying structural semantics of the input graph. Recent works usually sample negative samples from the same training batch with the positive samples or from an external irrelevant graph. However, a significant limitation lies in such strategies: the unavoidable problem of sampling false negative samples. In this paper, we propose a novel method to utilize Counterfactual mechanism to generate artificial hard negative samples for Graph Contrastive learning, namely CGC. We utilize a counterfactual mechanism to produce hard negative samples, ensuring that the generated samples are similar but have labels that differ from the positive sample. The proposed method achieves satisfying results on several datasets. It outperforms some traditional unsupervised graph learning methods and some SOTA graph contrastive learning methods. We also conducted some supplementary experiments to illustrate the proposed method, including the performances of CGC with different hard negative samples and evaluations for hard negative samples generated with different similarity measurements. The implementation code is available online to ease reproducibility1.Keywords
All Related Versions
Funding Information
- Huawei Innovation Research Program
- NSF (National Science Foundation) (III-1763325, III-1909323, III-2106758, SaTC-1930941)
- HKIDS Early Career Research Grant (9360163)
- Australian Research Council (DP220103717, DP200101374, LP170100891, LE220100078)
- Ant Group (CCF-Ant Research Fund)
- APRC - CityU New Research Initiatives (9610565)
- SIRG - CityU Strategic Interdisciplinary Research Grant (7020046, 7020074)
This publication has 13 references indexed in Scilit:
- Graph Trend Filtering Networks for Recommendation2022
- Counterfactual Explanations in Explainable AI: A Tutorial2021
- Graph Contrastive Learning with Adaptive Augmentation2021
- GCC2020
- Momentum Contrast for Unsupervised Visual Representation Learning2020
- A Survey of Data-driven and Knowledge-aware eXplainable AIIEEE Transactions on Knowledge and Data Engineering, 2020
- Sub2Vec: Feature Learning for Subgraphs2018
- Unsupervised Feature Learning via Non-parametric Instance Discrimination2018
- Faster Kernels for Graphs with Continuous Attributes via Hashing2016
- Protein function prediction via graph kernelsBioinformatics, 2005