Abstract     Description     Claims     Drawing  

On pixel-wise explanations for non-linear classifier decisions by layer-wise relevance propagation   [0004] 
Top-down neural attention by excitation backprop   [0005]  [0075] 
Look and think twice: Capturing top-down visual attention with feedback convolutional neural networks   [0006] 
Very Deep Convolutional Networks for Large-Scale Image Recognition   [0035] 
Going deeper with convolutions   [0035] 
Deep residual learning for image recognition   [0035] 
Very deep convolutional networks for large-scale image recognition   [0077]