Explaining a series of models by propagating Shapley values.

Explaining a series of models by propagating Shapley values.
复制标题

通过传播Shapley值来解释一系列模型。

DOI:
10.1038/s41467-022-31384-3
复制
发表时间:
2022-08-03
影响因子:
16.6
通讯作者:
Lee, Su-In
Lee, Su-In
中科院分区:
综合性期刊1区
文献类型:
--
作者:
Chen, Hugh;Lundberg, Scott M.;Lee, Su-In

文献摘要

参考文献

被引文献

相似文献

局部特征归因方法越来越多地用于解释复杂的机器学习模型。然而,目前的方法是有限的,因为它们是非常昂贵的计算或无法解释一系列分布式模型,其中每个模型是由一个单独的机构拥有。后者尤其重要,因为它经常出现在要求解释的金融领域。在这里,我们提出了广义DeepSHAP(G-DeepSHAP),这是一种易于处理的方法,可以基于与Shapley值的连接,通过复杂的一系列模型来传播局部特征属性。我们在生物、健康和金融数据集上评估了G-DeepSHAP,以表明它比现有的模型不可知归因技术更快地提供了同样突出的解释,并展示了它在一系列重要的分布式模型设置中的应用。与生物学、医学和金融领域的任务相关的一系列机器学习模型通常涉及复杂的特征归因技术。作者介绍了一种易于处理的方法来计算一系列机器学习模型的局部特征属性,这些模型受到Shapley值连接的启发。
Local feature attribution methods are increasingly used to explain complex machine learning models. However, current methods are limited because they are extremely expensive to compute or are not capable of explaining a distributed series of models where each model is owned by a separate institution. The latter is particularly important because it often arises in finance where explanations are mandated. Here, we present Generalized DeepSHAP (G-DeepSHAP), a tractable method to propagate local feature attributions through complex series of models based on a connection to the Shapley value. We evaluate G-DeepSHAP across biological, health, and financial datasets to show that it provides equally salient explanations an order of magnitude faster than existing model-agnostic attribution techniques and demonstrate its use in an important distributed series of models setting. Series of machine learning models, relevant for tasks in biology, medicine, and finance, usually involve complex feature attribution techniques. The authors introduce a tractable method to compute local feature attributions for a series of machine learning models inspired by connections to the Shapley value.
DOI: 10.3233/jad-179939
发表时间: 2018
期刊: Journal of Alzheimer's disease : JAD
影响因子: --
作者:
Bennett DA;Buchman AS;Boyle PA;Barnes LL;Wilson RS;Schneider JA
通讯作者: Schneider JA
DOI: 10.1609/aimag.v38i3.2741
发表时间: 2017-09-01
期刊: AI MAGAZINE
影响因子: 0.9
作者:
Goodman, Bryce;Flaxman, Seth
通讯作者: Flaxman, Seth
DOI: 10.1016/j.artint.2021.103502
发表时间: 2021-04-19
影响因子: 14.4
作者:
Aas, Kjersti;Jullum, Martin;Loland, Anders
通讯作者: Loland, Anders
DOI: 10.1038/nature10983
发表时间: 2012-04-18
期刊: NATURE
影响因子: 64.8
作者:
Curtis, Christina;Shah, Sohrab P.;Chin, Suet-Feung;Turashvili, Gulisa;Rueda, Oscar M.;Dunning, Mark J.;Speed, Doug;Lynch, Andy G.;Samarajiwa, Shamith;Yuan, Yinyin;Graef, Stefan;Ha, Gavin;Haffari, Gholamreza;Bashashati, Ali;Russell, Roslin;McKinney, Steven;Langerod, Anita;Green, Andrew;Provenzano, Elena;Wishart, Gordon;Pinder, Sarah;Watson, Peter;Markowetz, Florian;Murphy, Leigh;Ellis, Ian;Purushotham, Arnie;Borresen-Dale, Anne-Lise;Brenton, James D.;Tavare, Simon;Caldas, Carlos;Aparicio, Samuel
通讯作者: Aparicio, Samuel
DOI: 10.1098/rsif.2017.0520
发表时间: 2017-09-01
影响因子: 3.9
作者:
Bhatt, Samir;Cameron, Ewan;Gething, Peter W.
通讯作者: Gething, Peter W.