A causal framework for explaining the predictions of black-box sequence-to-sequence models

David Alvarez-Melis, Tommi Jaakkola


Abstract
We interpret the predictions of any black-box structured input-structured output model around a specific input-output pair. Our method returns an “explanation” consisting of groups of input-output tokens that are causally related. These dependencies are inferred by querying the model with perturbed inputs, generating a graph over tokens from the responses, and solving a partitioning problem to select the most relevant components. We focus the general approach on sequence-to-sequence problems, adopting a variational autoencoder to yield meaningful input perturbations. We test our method across several NLP sequence generation tasks.
Anthology ID:
D17-1042
Volume:
Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing
Month:
September
Year:
2017
Address:
Copenhagen, Denmark
Venue:
EMNLP
SIG:
SIGDAT
Publisher:
Association for Computational Linguistics
Note:
Pages:
412–421
Language:
URL:
https://www.aclweb.org/anthology/D17-1042
DOI:
10.18653/v1/D17-1042
Bib Export formats:
BibTeX MODS XML EndNote
PDF:
http://aclanthology.lst.uni-saarland.de/D17-1042.pdf
Attachment:
 D17-1042.Attachment.zip