Top-down Neural Attention by Excitation Backprop
International Journal of Computer Vision, Volume 126, Issue 10, 2017, Pages 1084-1102.
We aim to model the top-down attention of a convolutional neural network (CNN) classifier for generating task-specific attention maps. Inspired by a top-down human visual attention model, we propose a new backpropagation scheme, called Excitation Backprop, to pass along top-down signals downwards in the network hierarchy via a probabilist...More
PPT (Upload PPT)