Learning Groupwise Explanations for Black-Box Models

Gao, Jingyue; Wang, Xiting; Wang, Yasha; Yan, Yulan; Xie, Xing

doi:10.24963/ijcai.2021/330

Learning Groupwise Explanations for Black-Box Models

Jingyue Gao, Xiting Wang, Yasha Wang, Yulan Yan, Xing Xie

Download BibTeX Cite

Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence. Main Track. Pages 2396-2402.

https://doi.org/10.24963/ijcai.2021/330

Keywords

Machine Learning: Explainable/Interpretable Machine Learning

Abstract

We study two user demands that are important during the exploitation of explanations in practice: 1) understanding the overall model behavior faithfully with limited cognitive load and 2) predicting the model behavior accurately on unseen instances. We illustrate that the two user demands correspond to two major sub-processes in the human cognitive process and propose a unified framework to fulfill them simultaneously. Given a local explanation method, our framework jointly 1) learns a limited number of groupwise explanations that interpret the model behavior on most instances with high fidelity and 2) specifies the region where each explanation applies. Experiments on six datasets demonstrate the effectiveness of our method.

Jingyue Gao, Xiting Wang, Yasha Wang, Yulan Yan, and Xing Xie. 2021. Learning Groupwise Explanations for Black-Box Models. In Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, IJCAI-21. IJCAI, 2396–2402. https://doi.org/10.24963/ijcai.2021/330

@inproceedings{ijcai2021p330,
  title     = {Learning Groupwise Explanations for Black-Box Models},
  author    = {Gao, Jingyue and Wang, Xiting and Wang, Yasha and Yan, Yulan and Xie, Xing},
  booktitle = {Proceedings of the Thirtieth International Joint Conference on
               Artificial Intelligence, {IJCAI-21}},
  publisher = {International Joint Conferences on Artificial Intelligence Organization},
  editor    = {Prof. Zhi-Hua Zhou},
  pages     = {2396--2402},
  year      = {2021},
  month     = {8},
  note      = {Main Track},
  doi       = {10.24963/ijcai.2021/330},
  url       = {https://doi.org/10.24963/ijcai.2021/330},
}

TY  - CPAPER
TI  - Learning Groupwise Explanations for Black-Box Models
AU  - Gao, Jingyue
AU  - Wang, Xiting
AU  - Wang, Yasha
AU  - Yan, Yulan
AU  - Xie, Xing
T2  - Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, IJCAI-21
PB  - International Joint Conferences on Artificial Intelligence Organization
PY  - 2021
SP  - 2396
EP  - 2402
DO  - 10.24963/ijcai.2021/330
UR  - https://doi.org/10.24963/ijcai.2021/330
ER  -