Attention vs non-attention for a Shapley-based explanation method

TL;DR

This paper extends a Shapley-based explanation method, Contextual Decomposition, to attention models in NLP, comparing their syntactic processing and demonstrating its effectiveness for attention-based neural networks.

Contribution

We adapt Contextual Decomposition for attention models and compare their processing of syntactic structures in English and Dutch, revealing differences from non-attention models.

Findings

CD is effective for attention-based models

Attention and non-attention models process syntax differently

Similar processing behavior observed across languages

Abstract

The field of explainable AI has recently seen an explosion in the number of explanation methods for highly non-linear deep neural networks. The extent to which such methods -- that are often proposed and tested in the domain of computer vision -- are appropriate to address the explainability challenges in NLP is yet relatively unexplored. In this work, we consider Contextual Decomposition (CD) -- a Shapley-based input feature attribution method that has been shown to work well for recurrent NLP models -- and we test the extent to which it is useful for models that contain attention operations. To this end, we extend CD to cover the operations necessary for attention-based models. We then compare how long distance subject-verb relationships are processed by models with and without attention, considering a number of different syntactic structures in two different languages: English and Dutch. Our experiments confirm that CD can successfully be applied for attention-based models as well, providing an alternative Shapley-based attribution method for modern neural networks. In particular, using CD, we show that the English and Dutch models demonstrate similar processing behaviour, but that under the hood there are consistent differences between our attention and non-attention models.