Regularization, Semi-supervision, and Supervision for a Plausible Attention-Based Explanation

TL;DR

This paper introduces three new constraints to improve the plausibility of attention maps in NLP models, enhancing interpretability without significantly harming classification performance.

Contribution

It proposes regularization, semi-supervision, and human supervision techniques to make attention maps more plausible as explanations in NLP models.

Findings

All proposed techniques improve attention map plausibility.

Specific human annotation instructions may negatively impact classification accuracy.

The contextualization layer is key to identifying plausible tokens.

Abstract

Attention mechanism is contributing to the majority of recent advances in machine learning for natural language processing. Additionally, it results in an attention map that shows the proportional influence of each input in its decision. Empirical studies postulate that attention maps can be provided as an explanation for model output. However, it is still questionable to ask whether this explanation helps regular people to understand and accept the model output (the plausibility of the explanation). Recent studies show that attention weights in the RNN encoders are hardly plausible because they spread on input tokens. We thus propose 3 additional constraints to the learning objective function to improve the plausibility of the attention map: regularization to increase the attention weight sparsity, semi-supervision to supervise the map by a heuristic and supervision by human annotation. Results show that all techniques can improve the attention map plausibility at some level. We also observe that specific instructions for human annotation might have a negative effect on classification performance. Beyond the attention map, the result of experiments on text classification tasks also shows that no matter how the constraint brings the gain, the contextualization layer plays a crucial role in finding the right space for finding plausible tokens.