Abstract
Interpretability methods for neural networks are difficult to evaluate because we do not understand the black-box models typically used to test them. This paper proposes a framework in which interpretability methods are evaluated using manually constructed networks, which we call white-box networks, whose behavior is understood a priori. We evaluate five methods for producing attribution heatmaps by applying them to white-box LSTM classifiers for tasks based on formal languages. Although our white-box classifiers solve their tasks perfectly and transparently, we find that all five attribution methods fail to produce the expected model explanations.- Anthology ID:
- 2020.blackboxnlp-1.28
- Volume:
- Proceedings of the Third BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP
- Month:
- November
- Year:
- 2020
- Address:
- Online
- Editors:
- Afra Alishahi, Yonatan Belinkov, Grzegorz Chrupała, Dieuwke Hupkes, Yuval Pinter, Hassan Sajjad
- Venue:
- BlackboxNLP
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 300–313
- Language:
- URL:
- https://aclanthology.org/2020.blackboxnlp-1.28
- DOI:
- 10.18653/v1/2020.blackboxnlp-1.28
- Cite (ACL):
- Yiding Hao. 2020. Evaluating Attribution Methods using White-Box LSTMs. In Proceedings of the Third BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP, pages 300–313, Online. Association for Computational Linguistics.
- Cite (Informal):
- Evaluating Attribution Methods using White-Box LSTMs (Hao, BlackboxNLP 2020)
- PDF:
- https://preview.aclanthology.org/nschneid-patch-1/2020.blackboxnlp-1.28.pdf
- Code
- yidinghao/whitebox-lstm