Data Augmentation for Graph Classification

TL;DR

This paper introduces graph data augmentation techniques and a model evolution framework to improve graph classification accuracy on small datasets, reducing overfitting and enhancing model performance.

Contribution

It presents two heuristic algorithms for graph augmentation and a generic framework, M-Evolve, for iterative model improvement on limited data.

Findings

M-Evolve improves accuracy by 3-12% on benchmark datasets.

Graph augmentation reduces overfitting in small-scale datasets.

The approach is effective across multiple graph classification tasks.

Abstract

Graph classification, which aims to identify the category labels of graphs, plays a significant role in drug classification, toxicity detection, protein analysis etc. However, the limitation of scale of benchmark datasets makes it easy for graph classification models to fall into over-fitting and undergeneralization. Towards this, we introduce data augmentation on graphs and present two heuristic algorithms: random mapping and motif-similarity mapping, to generate more weakly labeled data for small-scale benchmark datasets via heuristic modification of graph structures. Furthermore, we propose a generic model evolution framework, M-Evolve, which combines graph augmentation, data filtration and model retraining to optimize pre-trained graph classifiers. Experiments conducted on six benchmark datasets demonstrate that M-Evolve helps existing graph classification models alleviate over-fitting when training on small-scale benchmark datasets and yields an average improvement of 3-12% accuracy on graph classification tasks.