Saccader: Improving Accuracy of Hard Attention Models for Vision
This addresses the problem of interpretability in vision models for researchers and practitioners, though it is incremental in scaling hard attention to complex datasets.
The paper tackles the challenge of training hard attention models for interpretable image classification by proposing Saccader, which uses a pretraining step to initialize attention locations, achieving 75% top-1 and 91% top-5 accuracy on ImageNet while attending to less than one-third of the image.
Although deep convolutional neural networks achieve state-of-the-art performance across nearly all image classification tasks, their decisions are difficult to interpret. One approach that offers some level of interpretability by design is \textit{hard attention}, which uses only relevant portions of the image. However, training hard attention models with only class label supervision is challenging, and hard attention has proved difficult to scale to complex datasets. Here, we propose a novel hard attention model, which we term Saccader. Key to Saccader is a pretraining step that requires only class labels and provides initial attention locations for policy gradient optimization. Our best models narrow the gap to common ImageNet baselines, achieving $75\%$ top-1 and $91\%$ top-5 while attending to less than one-third of the image.