arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2202.00734cs.LGstat.ML
收藏

地方政府诚信评价框架研究

Framework for Evaluating Faithfulness of Local Explanations

University of California San Diego(加利福尼亚大学圣地亚哥分校);Tel Aviv University(特拉维夫大学)

Sanjoy Dasgupta, Nave Frost, Michal Moshkovitz

摘要:

我们研究了解释系统对潜在预测模型的忠诚度。我们表明,这可以被捕获的两个属性,一致性和充分性,并引入定量措施的程度,这些举行。有趣的是,这些措施取决于测试时的数据分布。对于各种现有的解释系统,如锚,我们分析研究这些量。我们还提供了估计和样本的复杂性界限,经验确定的忠实性的黑盒解释系统。最后,我们通过实验验证了新的性质和估计。

英文摘要:

We study the faithfulness of an explanation system to the underlying prediction model. We show that this can be captured by two properties, consistency and sufficiency, and introduce quantitative measures of the extent to which these hold. Interestingly, these measures depend on the test-time data distribution. For a variety of existing explanation systems, such as anchors, we analytically study these quantities. We also provide estimators and sample complexity bounds for empirically determining the faithfulness of black-box explanation systems. Finally, we experimentally validate the new properties and estimators.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑