Achieving Explainability in a Visual Hard Attention Model through Content Prediction

Samrudhdhi B. Rangrej, James J. Clark

Keywords: visual hard attention, glimpses, explainability, bayesian optimal experiment design, variational autoencoder

Abstract: A visual hard attention model actively selects and observes a sequence of subregions in an image to make a prediction. Unlike in the deep convolution network, in hard attention it is explainable which regions of the image contributed to the prediction. However, the attention policy used by the model to select these regions is not explainable. The majority of hard attention models determine the attention-worthy regions by first analyzing a complete image. However, it may be the case that the entire image is not available in the beginning but instead sensed gradually through a series of partial observations. In this paper, we design an efficient hard attention model for classifying partially observable scenes. The attention policy used by our model is explainable and non-parametric. The model estimates expected information gain (EIG) obtained from attending various regions by predicting their content ahead of time. It compares EIG using Bayesian Optimal Experiment Design and attends to the region with maximum EIG. We train our model with a differentiable objective, optimized using gradient descent, and test it on several datasets. The performance of our model is comparable to or better than the baseline models.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

One-sentence Summary: The hard attention model learns explainable attention policy by predicting the content of the glimpses and using it to find an optimal location to attend.

Reviewed Version (pdf): https://openreview.net/references/pdf?id=m8GPVdyyPb