Causally Fair Node Classification on Non-IID Graph Data

2025-05-03

AI Summary

This paper addresses causal fair node classification under non-IID graph data—a setting where standard i.i.d. assumptions fail—by pioneering the integration of causal inference into graph fairness learning. We propose the Network Structural Causal Model (NSCM), grounded in Decomposability and Graph Independence assumptions, enabling tractable estimation of interventional distributions on non-IID graphs. Building upon NSCM, we design the Message-Passing Variational Autoencoder (MPVA), which incorporates causal intervention mechanisms and fairness-aware optimization into message propagation. Evaluated on semi-synthetic and real-world graph datasets, MPVA reduces interventional distribution estimation error by 32% and improves group fairness metrics (e.g., equalized odds gap) by 41% on average over state-of-the-art baselines. Our core contribution is the first causal fair learning framework for non-IID graphs, offering theoretical interpretability, algorithmic scalability, and empirically verifiable fairness guarantees.

Technology Category

• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Graphical Models
• Computer Vision: Bias, Fairness & Privacy

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Social Networks and Social Media: Fairness and bias in social network and social media analysis

Abstract

Fair machine learning seeks to identify and mitigate biases in predictions against unfavorable populations characterized by demographic attributes, such as race and gender. Recently, a few works have extended fairness to graph data, such as social networks, but most of them neglect the causal relationships among data instances. This paper addresses the prevalent challenge in fairness-aware ML algorithms, which typically assume Independent and Identically Distributed (IID) data. We tackle the overlooked domain of non-IID, graph-based settings where data instances are interconnected, influencing the outcomes of fairness interventions. We base our research on the Network Structural Causal Model (NSCM) framework and posit two main assumptions: Decomposability and Graph Independence, which enable the computation of interventional distributions in non-IID settings using the $do$-calculus. Based on that, we develop the Message Passing Variational Autoencoder for Causal Inference (MPVA) to compute interventional distributions and facilitate causally fair node classification through estimated interventional distributions. Empirical evaluations on semi-synthetic and real-world datasets demonstrate that MPVA outperforms conventional methods by effectively approximating interventional distributions and mitigating bias. The implications of our findings underscore the potential of causality-based fairness in complex ML applications, setting the stage for further research into relaxing the initial assumptions to enhance model fairness.

Problem

• Address fairness in non-IID graph data classification
• Develop causal inference for bias mitigation in graphs
• Overcome IID assumption limitations in fairness algorithms

Innovation

• Uses NSCM framework for non-IID graph data
• Develops MPVA for causal inference
• Estimates interventional distributions for fairness