説明可能AI(XAI) DeepExplain,DeepLIFT,DeepSHAP

SHAPでの説明をするツールを紹介したい。

シュープレイ値について(Wikipedia)

https://en.wikipedia.org/wiki/Shapley_value#In_machine_learning

SHAP (SHapley Additive exPlanations) DeepSHAP

https://github.com/slundberg/shap

このツールは

・Tree ensemble example (XGBoost/LightGBM/CatBoost/scikit-learn/pyspark models)

・Natural language example (transformers)

・Deep learning example with DeepExplainer (TensorFlow/Keras models)

・Deep learning example with GradientExplainer (TensorFlow/Keras/PyTorch models)

・Model agnostic example with KernelExplainer (explains any function)

・SHAP Interaction Values

wikipediaでの参考文献

SHAP値のmachine learningへの導入のきっかけとなった論文

 Lundberg, Scott M.; Lee, Su-In. “A Unified Approach to Interpreting Model Predictions”Advances in Neural Information Processing Systems30: 4765–4774. Retrieved 2021-01-30.

https://papers.nips.cc/paper/2017/hash/8a20a8621978632d76c43dfd28b67767-Abstract.html

・Locally Interpretable Model-Agnostic Explanations (LIME)

Ribeiro, Marco Tulio; Singh, Sameer; Guestrin, Carlos (2016-08-13). “Why Should I Trust You?”. New York, NY, USA: ACM.

DeepLIFT

https://github.com/kundajelab/deeplift

Shrikumar, Avanti; Greenside, Peyton; Kundaje, Anshul (2017-07-17). “Learning Important Features Through Propagating Activation Differences”PMLR. pp. 3145–3153. ISSN 2640-3498. Retrieved 2021-01-30.

Layer-Wise Relevance Propagation

Bach, Sebastian; Binder, Alexander; Montavon, Grégoire; Klauschen, Frederick; Müller, Klaus-Robert; Samek, Wojciech (2015-07-10). Suarez, Oscar Deniz (ed.). “On Pixel-Wise Explanations for Non-Linear Classifier Decisions by Layer-Wise Relevance Propagation”PLOS ONE. Public Library of Science (PLoS). 10 (7): e0130140. doi:10.1371/journal.pone.0130140ISSN 1932-6203.

・DeepExplain

https://github.com/marcoancona/DeepExplain

経緯としてはICML2018でAnconaらのDeepExplainを契機にして、DeepLIFTとDeepSHAPが生まれたらしい。

どちらかの違いについてはDeepLIFTのFAQに詳しい。結論としてはまずDeepSHAPでやり、データの規模が大きい場合はDeepLIFTがおすすめらしい。

https://github.com/kundajelab/deeplift#what-are-the-similarities-and-differences-between-the-deeplift-like-implementations-in-deepexplain-from-ancona-et-al-iclr-2018-and-deepshapdeepexplainer-from-the-shap-repository

SHAPのメソッドを試した記事

・https://qiita.com/shin_mura/items/cde01198552eda9146b7

・https://dropout009.hatenablog.com/entry/2019/11/20/091450?_ga=2.185612329.1656579419.1574324138-1185090719.1545096259

最新情報をチェックしよう!