Unifying Perspectives: Plausible Counterfactual Explanations on Global, Group-wise, and Local Levels

2024-05-27

arXiv.org

AI Summary

Existing XAI research lacks a unified framework for jointly generating local, global, and group-wise counterfactual explanations (CFs), particularly failing to model feasibility and jointly optimize group-wise CFs (GWCFs). Method: We propose the first single-stage gradient-based optimization framework that tightly integrates dynamic clustering–driven group partitioning with multi-objective CF generation. It explicitly enforces data manifold feasibility constraints and jointly optimizes for effectiveness, proximity, and plausibility via differentiable optimization and a multi-term loss (L2 distance, confidence margin, and feasibility regularization). Contribution/Results: On multiple benchmark datasets, our method improves GWCF effectiveness by 23.6%, reduces average proximity by 18.4%, and achieves a feasibility score of 0.91. A user study confirms significantly superior interpretability and decision-support utility compared to two-stage baselines.

Technology Category

• Humans and AI: Explainable AI (XAI) for Human Understanding
• Search and Optimization: Local Search
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI

Abstract

The growing complexity of AI systems has intensified the need for transparency through Explainable AI (XAI). Counterfactual explanations (CFs) offer actionable"what-if"scenarios on three levels: Local CFs providing instance-specific insights, Global CFs addressing broader trends, and Group-wise CFs (GWCFs) striking a balance and revealing patterns within cohesive groups. Despite the availability of methods for each granularity level, the field lacks a unified method that integrates these complementary approaches. We address this limitation by proposing a gradient-based optimization method for differentiable models that generates Local, Global, and Group-wise Counterfactual Explanations in a unified manner. We especially enhance GWCF generation by combining instance grouping and counterfactual generation into a single efficient process, replacing traditional two-step methods. Moreover, to ensure trustworthiness, we innovatively introduce the integration of plausibility criteria into the GWCF domain, making explanations both valid and realistic. Our results demonstrate the method's effectiveness in balancing validity, proximity, and plausibility while optimizing group granularity, with practical utility validated through practical use cases.

Problem

• Lack of unified method for Local, Global, Group-wise Counterfactual Explanations
• Traditional GWCF generation uses inefficient two-step process
• Need plausibility criteria for trustworthy Group-wise Counterfactual Explanations

Innovation

• Gradient-based unified counterfactual explanation generation
• Single-process group-wise CFs with plausibility
• Integrates local, global, group-wise explanations efficiently