Evaluating Explanation Without Ground Truth in Interpretable Machine Learning
Evaluating Explanation Without Ground Truth in Interpretable Machine Learning
复制标题
在可解释的机器学习中评估没有基本事实的解释
DOI：
--
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
Xia Hu
中科院分区：
文献类型：
--
作者：
Fan Yang;Mengnan Du;Xia Hu
可解释的机器学习（IML）在许多现实应用中变得越来越重要，例如自动驾驶汽车和医疗诊断，在这些应用中，解释是非常受欢迎的，可以帮助人们更好地理解机器学习系统如何工作，并进一步增强他们对系统的信任。然而，由于场景的多样化和解释的主观性，我们很少有 IML 中生成的解释质量的基准评估的基本事实。具有解释质量的意识不仅对于评估系统边界很重要，而且还有助于在实际环境中实现人类用户的真正好处。为了对 IML 的评估进行基准测试，在本文中，我们严格定义了评估解释的问题，并系统地回顾了现有技术的成果。具体来说，我们用正式的定义总结了解释的三个一般方面（即普遍性、保真度和说服力），并分别回顾了不同任务下每个方面的代表性方法。此外，根据开发者和最终用户的分层需求设计了统一的评估框架，可以在实践中轻松应用于不同场景。最后，讨论了未解决的问题，并提出了当前评估技术的一些局限性以供未来的探索。
Interpretable Machine Learning (IML) has become increasingly important in many real-world applications, such as autonomous cars and medical diagnosis, where explanations are significantly preferred to help people better understand how machine learning systems work and further enhance their trust towards systems. However, due to the diversified scenarios and subjective nature of explanations, we rarely have the ground truth for benchmark evaluation in IML on the quality of generated explanations. Having a sense of explanation quality not only matters for assessing system boundaries, but also helps to realize the true benefits to human users in practical settings. To benchmark the evaluation in IML, in this article, we rigorously define the problem of evaluating explanations, and systematically review the existing efforts from state-of-the-arts. Specifically, we summarize three general aspects of explanation (i.e., generalizability, fidelity and persuasibility) with formal definitions, and respectively review the representative methodologies for each of them under different tasks. Further, a unified evaluation framework is designed according to the hierarchical needs from developers and end-users, which could be easily adopted for different scenarios in practice. In the end, open problems are discussed, and several limitations of current evaluation techniques are raised for future explorations.
可解释的主动学习
DOI：
--
发表时间：
2018
期刊：
and Transparency
影响因子：
--
作者：
Phillips, Richard;Chang, Kyu Hyun;Friedler, Sorelle A.
通讯作者：
Friedler, Sorelle A.
DOI：
10.1073/pnas.1900654116
发表时间：
2019-10-29
影响因子：
11.1
作者：
Murdoch, W. James;Singh, Chandan;Yu, Bin
通讯作者：
Yu, Bin
影响因子：
2.9
作者：
Montavon, Gregoire;Samek, Wojciech;Mueller, Klaus-Robert
通讯作者：
Mueller, Klaus-Robert