Sparse Causal Discovery with Generative Intervention for Unsupervised Graph Domain Adaptation

2025-07-10

AI Summary

Unsupervised Graph Domain Adaptation (UGDA) faces three key challenges: entanglement of causal and spurious features, failure of global distribution alignment, and error accumulation in pseudo-labeling. To address these, we propose a sparse causal discovery–driven generative intervention framework. First, we construct a sparse causal graph to model stable inter-node causal relationships, thereby disentangling causal features and breaking local spurious correlations. Second, we design a class-adaptive dynamic intervention mechanism that integrates mutual information bottleneck principles with variational inference for robust feature reconfiguration. Third, we introduce covariance consistency constraints and a dynamic pseudo-label calibration strategy to suppress error propagation. Extensive experiments on multiple real-world graph benchmarks demonstrate significant improvements over state-of-the-art methods, validating the effectiveness and stability of our approach for cross-domain graph representation transfer.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Reasoning under Uncertainty: Graphical Models
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies

Abstract

Unsupervised Graph Domain Adaptation (UGDA) leverages labeled source domain graphs to achieve effective performance in unlabeled target domains despite distribution shifts. However, existing methods often yield suboptimal results due to the entanglement of causal-spurious features and the failure of global alignment strategies. We propose SLOGAN (Sparse Causal Discovery with Generative Intervention), a novel approach that achieves stable graph representation transfer through sparse causal modeling and dynamic intervention mechanisms. Specifically, SLOGAN first constructs a sparse causal graph structure, leveraging mutual information bottleneck constraints to disentangle sparse, stable causal features while compressing domain-dependent spurious correlations through variational inference. To address residual spurious correlations, we innovatively design a generative intervention mechanism that breaks local spurious couplings through cross-domain feature recombination while maintaining causal feature semantic consistency via covariance constraints. Furthermore, to mitigate error accumulation in target domain pseudo-labels, we introduce a category-adaptive dynamic calibration strategy, ensuring stable discriminative learning. Extensive experiments on multiple real-world datasets demonstrate that SLOGAN significantly outperforms existing baselines.

Problem

• Disentangle causal-spurious features in graph domain adaptation
• Achieve stable graph representation transfer via sparse modeling
• Mitigate error accumulation in target domain pseudo-labels

Innovation

• Sparse causal modeling for stable graph transfer
• Generative intervention breaks spurious correlations
• Dynamic calibration ensures stable discriminative learning