[论文解读] Explainable Artificial Intelligence (XAI): Concepts, Taxonomies, Opportunities and Challenges toward Responsible AI

Alejandro Barredo Arrieta, Natalia Díaz-Rodríguez+10 more|arXiv (Cornell University)|Oct 22, 2019

Explainable Artificial Intelligence (XAI)被引用 34

一句话总结

一份全面综述，定义 XAI 中的可解释性，以受众为中心的解读，提出透明和事后方法的分类法，分析挑战，并讨论负责任 AI（Responsible AI）包括隐私、公平和数据融合的影响。

ABSTRACT

In the last years, Artificial Intelligence (AI) has achieved a notable momentum that may deliver the best of expectations over many application sectors across the field. For this to occur, the entire community stands in front of the barrier of explainability, an inherent problem of AI techniques brought by sub-symbolism (e.g. ensembles or Deep Neural Networks) that were not present in the last hype of AI. Paradigms underlying this problem fall within the so-called eXplainable AI (XAI) field, which is acknowledged as a crucial feature for the practical deployment of AI models. This overview examines the existing literature in the field of XAI, including a prospect toward what is yet to be reached. We summarize previous efforts to define explainability in Machine Learning, establishing a novel definition that covers prior conceptual propositions with a major focus on the audience for which explainability is sought. We then propose and discuss about a taxonomy of recent contributions related to the explainability of different Machine Learning models, including those aimed at Deep Learning methods for which a second taxonomy is built. This literature analysis serves as the background for a series of challenges faced by XAI, such as the crossroads between data fusion and explainability. Our prospects lead toward the concept of Responsible Artificial Intelligence, namely, a methodology for the large-scale implementation of AI methods in real organizations with fairness, model explainability and accountability at its core. Our ultimate goal is to provide newcomers to XAI with a reference material in order to stimulate future research advances, but also to encourage experts and professionals from other disciplines to embrace the benefits of AI in their activity sectors, without any prior bias for its lack of interpretability.

研究动机与目标

• 提供一个统一的、面向受众的 ML/AI 中可解释性的定义。
• 为 XAI 制定覆盖透明模型与事后解释的分类法，包括深度学习相关方法。
• 调研文献以识别 XAI 与数据融合中的目标、方法与挑战。
• 讨论在负责任 AI 框架下对隐私、鲁棒性和保密性的影响。
• 勾勒未来方向和研究需求，以推动切实可行、负责任的 AI 部署。

提出的方法

• 以面向受众的可理解性为基础来界定可解释性的定义。
• 将 ML 模型分类为透明（按设计）或可通过事后技术解释的模型，并设定透明度层级（算法透明、可分解性、可模拟性）。
• 构建两个分类法：一個用于普遍 ML 模型（包括深度与非深度），一個专门用于深度学习解释（逐层、表征、注意机制）。
• 审阅约400 条贡献以识别目标与方法，并将其综合为一个连贯的框架。
• 在负责任 AI 的范式下，讨论 XAI 与数据融合、隐私和安全之间的相互作用。

实验结果

研究问题

• RQ1 如何在考虑受众和目的的前提下定义可解释性？
• RQ2 现有的 XAI 方法对透明和事后方法有哪些分类和分法？
• RQ3 推动 XAI 的目标在各学科中为何，以及它们如何映射到受众和应用？
• RQ4 在评估可解释性以及在深度学习和数据融合情境中应用 XAI 时，仍存在哪些挑战？
• RQ5 XAI 如何促进包括公平、问责和隐私在内的更广泛的负责任 AI 概念？

主要发现

• 提出一个新颖的、面向受众的可解释性定义，强调提供给特定受众的细节和原因。
• 开发两套全面的分类法：1）可解释性（透明设计与事后）以及 2）深度学习 XAI 方法，且将标准与 DL 特征关联。
• 识别一系列广泛的 XAI 目标（可信性、因果性、可迁移性、信息性、置信、公平、可访问性、互动性、隐私意识），并将其映射到目标受众。
• 强调没有单一普遍公认的可解释性定义，强调需要用于评估可解释性及其质量的度量指标。
• 讨论 XAI 与数据融合、隐私、对抗性鲁棒性以及负责任 AI 治理之间的挑战。
• 倡导将负责任 AI 作为一个框架，将可解释性置于公平与问责同等重要的地位，以促进现实世界部署。