当前位置: X-MOL 学术arXiv.cs.CY › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Counterfactual Explanations as Interventions in Latent Space
arXiv - CS - Computers and Society Pub Date : 2021-06-14 , DOI: arxiv-2106.07754
Riccardo Crupi, Alessandro Castelnovo, Daniele Regoli, Beatriz San Miguel Gonzalez

Explainable Artificial Intelligence (XAI) is a set of techniques that allows the understanding of both technical and non-technical aspects of Artificial Intelligence (AI) systems. XAI is crucial to help satisfying the increasingly important demand of \emph{trustworthy} Artificial Intelligence, characterized by fundamental characteristics such as respect of human autonomy, prevention of harm, transparency, accountability, etc. Within XAI techniques, counterfactual explanations aim to provide to end users a set of features (and their corresponding values) that need to be changed in order to achieve a desired outcome. Current approaches rarely take into account the feasibility of actions needed to achieve the proposed explanations, and in particular they fall short of considering the causal impact of such actions. In this paper, we present Counterfactual Explanations as Interventions in Latent Space (CEILS), a methodology to generate counterfactual explanations capturing by design the underlying causal relations from the data, and at the same time to provide feasible recommendations to reach the proposed profile. Moreover, our methodology has the advantage that it can be set on top of existing counterfactuals generator algorithms, thus minimising the complexity of imposing additional causal constrains. We demonstrate the effectiveness of our approach with a set of different experiments using synthetic and real datasets (including a proprietary dataset of the financial domain).

中文翻译:

作为潜在空间干预的反事实解释

可解释人工智能 (XAI) 是一组技术,可让您了解人工智能 (AI) 系统的技术和非技术方面。XAI 对于帮助满足 \emph {trustworthy} 人工智能日益重要的需求至关重要,其特点是尊重人类自主、防止伤害、透明、问责等基本特征。在 XAI 技术中,反事实解释旨在提供最终用户需要更改以实现预期结果的一组功能(及其相应的值)。当前的方法很少考虑实现所提出的解释所需的行动的可行性,特别是它们没有考虑这些行动的因果影响。在本文中,我们将反事实解释作为潜在空间的干预 (CEILS) 提出,这是一种通过设计从数据中捕获潜在因果关系来生成反事实解释的方法,同时提供可行的建议以达到建议的配置文件。此外,我们的方法的优势在于它可以设置在现有的反事实生成器算法之上,从而最大限度地减少施加额外因果约束的复杂性。我们通过使用合成和真实数据集(包括金融领域的专有数据集)的一组不同实验证明了我们方法的有效性。并同时提供可行的建议以达到建议的配置文件。此外,我们的方法的优势在于它可以设置在现有的反事实生成器算法之上,从而最大限度地减少施加额外因果约束的复杂性。我们通过使用合成和真实数据集(包括金融领域的专有数据集)的一组不同实验证明了我们方法的有效性。并同时提供可行的建议以达到建议的配置文件。此外,我们的方法的优势在于它可以设置在现有的反事实生成器算法之上,从而最大限度地减少施加额外因果约束的复杂性。我们通过使用合成和真实数据集(包括金融领域的专有数据集)的一组不同实验证明了我们方法的有效性。
更新日期:2021-06-16
down
wechat
bug