FairACE: Achieving Degree Fairness in Graph Neural Networks via Contrastive and Adversarial Group-Balanced Training

--Fairness has been a significant challenge in graph neural networks (GNNs) since degree biases often result in unequal prediction performance among nodes with varying degrees. Existing GNN models focus on prediction accuracy, frequently overlooking fairness across different degree groups. T o address this issue, we propose a novel GNN framework, namely Fairness-A ware Asymmetric Contrastive Ensemble (FairACE), which integrates asymmetric contrastive learning with adversarial training to improve degree fairness. FairACE captures one-hop local neighborhood information and two-hop monophily similarity to create fairer node representations and employs a degree fairness regulator to balance performance between high-degree and low-degree nodes. During model training, a novel group-balanced fairness loss is proposed to minimize classification disparities across degree groups. In addition, we also propose a novel fairness metric, the Accuracy Distribution Gap (ADG), which can quantitatively assess and ensure equitable performance across different degree-based node groups. Experimental results on both synthetic and real-world datasets demonstrate that FairACE significantly improves degree fairness metrics while maintaini ng competitive accuracy in comparison to the state-of-the-art GNN models. RAPH Neural Networks (GNNs) have emerged as a powerful class of methods for learning representations of graph-structured data. These networks typically operate within a message-passing paradigm, where each node iteratively gathers and processes information from its neighborhood nodes across several layers [1]. By combining both the attributes of nodes and the underlying structural information, GNNs can generate rich and comprehensive representations for each node in the graph.