Identifying and Correcting Label Noise for Robust GNNs via Influence Contradiction

TL;DR

This paper introduces ICGNN, a method that uses graph structure and influence contradiction scores to identify and correct noisy labels in GNNs, improving robustness and accuracy on benchmark datasets.

Contribution

The paper proposes a novel noise detection and correction framework for GNNs that leverages influence contradiction scores and graph diffusion, enhancing robustness against label noise.

Findings

ICGNN effectively detects noisy labels using influence contradiction scores.

The approach improves GNN robustness and accuracy on benchmark datasets.

Incorporating pseudo-labeling further enhances model performance.

Abstract

Graph Neural Networks (GNNs) have shown remarkable capabilities in learning from graph-structured data with various applications such as social analysis and bioinformatics. However, the presence of label noise in real scenarios poses a significant challenge in learning robust GNNs, and their effectiveness can be severely impacted when dealing with noisy labels on graphs, often stemming from annotation errors or inconsistencies. To address this, in this paper we propose a novel approach called ICGNN that harnesses the structure information of the graph to effectively alleviate the challenges posed by noisy labels. Specifically, we first design a novel noise indicator that measures the influence contradiction score (ICS) based on the graph diffusion matrix to quantify the credibility of nodes with clean labels, such that nodes with higher ICS values are more likely to be detected as having noisy labels. Then we leverage the Gaussian mixture model to precisely detect whether the label of a node is noisy or not. Additionally, we develop a soft strategy to combine the predictions from neighboring nodes on the graph to correct the detected noisy labels. At last, pseudo-labeling for abundant unlabeled nodes is incorporated to provide auxiliary supervision signals and guide the model optimization. Experiments on benchmark datasets show the superiority of our proposed approach.