Towards Label Position Bias in Graph Neural Networks
This addresses a specific bias issue in GNNs for semi-supervised node classification, which is incremental as it builds on known biases in GNNs.
The paper tackles the problem of label position bias in Graph Neural Networks (GNNs), where nodes closer to labeled nodes perform better, and proposes an optimization framework that outperforms baseline methods and significantly reduces this bias.
Graph Neural Networks (GNNs) have emerged as a powerful tool for semi-supervised node classification tasks. However, recent studies have revealed various biases in GNNs stemming from both node features and graph topology. In this work, we uncover a new bias - label position bias, which indicates that the node closer to the labeled nodes tends to perform better. We introduce a new metric, the Label Proximity Score, to quantify this bias, and find that it is closely related to performance disparities. To address the label position bias, we propose a novel optimization framework for learning a label position unbiased graph structure, which can be applied to existing GNNs. Extensive experiments demonstrate that our proposed method not only outperforms backbone methods but also significantly mitigates the issue of label position bias in GNNs.