无需重训练即可实时生成多样化反事实解释，支持灵活约束控制。
DiCoFlex: Model-agnostic diverse counterfactuals with flexible control
- 基于条件归一化流，单次前向传播生成多个反事实样本。
- 在标准数据集上验证，有效性、多样性、贴近度和约束满足度均优于现有方法。
- 适用于医疗、金融等敏感决策领域，用户可实时调整稀疏性与可操作性约束。
反事实解释在可解释人工智能中至关重要，能提供直观的人类可理解替代方案以阐明机器学习模型的决策。尽管如此，现有反事实生成方法通常需要持续访问预测模型，对每个实例进行计算密集型优化，并且在不重新训练的情况下难以适应用户自定义的新约束。本文提出DiCoFlex，一种新型的模型无关、条件生成框架，可在一次前向传播中生成多个多样化的反事实样本。通过仅使用标注数据训练的条件归一化流，DiCoFlex在推理时实现了用户驱动的约束定制，如稀疏性和可操作性。在标准基准数据集上的大量实验表明，DiCoFlex在有效性、多样性、接近性及约束遵循方面均优于现有方法，使其成为敏感决策领域中反事实生成的实用且可扩展的解决方案。
原文摘要 · Abstract (English)
Counterfactual explanations play a pivotal role in explainable artificial intelligence (XAI) by offering intuitive, human-understandable alternatives that elucidate machine learning model decisions. Despite their significance, existing methods for generating counterfactuals often require constant access to the predictive model, involve computationally intensive optimization for each instance and lack the flexibility to adapt to new user-defined constraints without retraining. In this paper, we propose DiCoFlex, a novel model-agnostic, conditional generative framework that produces multiple diverse counterfactuals in a single forward pass. Leveraging conditional normalizing flows trained solely on labeled data, DiCoFlex addresses key limitations by enabling real-time user-driven customization of constraints such as sparsity and actionability at inference time. Extensive experiments on standard benchmark datasets show that DiCoFlex outperforms existing methods in terms of validity, diversity, proximity, and constraint adherence, making it a practical and scalable solution for counterfactual generation in sensitive decision-making domains.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。