How do Decisions Emerge across Layers in Neural Models? Interpretation with Differentiable Masking

Nicola De Cao, Michael Sejr Schlichtkrull, Wilker Aziz, Ivan Titov


Abstract
Attribution methods assess the contribution of inputs to the model prediction. One way to do so is erasure: a subset of inputs is considered irrelevant if it can be removed without affecting the prediction. Though conceptually simple, erasure’s objective is intractable and approximate search remains expensive with modern deep NLP models. Erasure is also susceptible to the hindsight bias: the fact that an input can be dropped does not mean that the model ‘knows’ it can be dropped. The resulting pruning is over-aggressive and does not reflect how the model arrives at the prediction. To deal with these challenges, we introduce Differentiable Masking. DiffMask learns to mask-out subsets of the input while maintaining differentiability. The decision to include or disregard an input token is made with a simple model based on intermediate hidden layers of the analyzed model. First, this makes the approach efficient because we predict rather than search. Second, as with probing classifiers, this reveals what the network ‘knows’ at the corresponding layers. This lets us not only plot attribution heatmaps but also analyze how decisions are formed across network layers. We use DiffMask to study BERT models on sentiment classification and question answering.
Anthology ID:
2020.emnlp-main.262
Volume:
Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)
Month:
November
Year:
2020
Address:
Online
Editors:
Bonnie Webber, Trevor Cohn, Yulan He, Yang Liu
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
3243–3255
Language:
URL:
https://aclanthology.org/2020.emnlp-main.262
DOI:
10.18653/v1/2020.emnlp-main.262
Bibkey:
Cite (ACL):
Nicola De Cao, Michael Sejr Schlichtkrull, Wilker Aziz, and Ivan Titov. 2020. How do Decisions Emerge across Layers in Neural Models? Interpretation with Differentiable Masking. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 3243–3255, Online. Association for Computational Linguistics.
Cite (Informal):
How do Decisions Emerge across Layers in Neural Models? Interpretation with Differentiable Masking (De Cao et al., EMNLP 2020)
Copy Citation:
PDF:
https://aclanthology.org/2020.emnlp-main.262.pdf
Optional supplementary material:
 2020.emnlp-main.262.OptionalSupplementaryMaterial.zip
Video:
 https://slideslive.com/38938648
Code
 nicola-decao/diffmask +  additional community code
Data
SST