Implementation of Class Model Visualisation as described in: Simonyan, K., Vedaldi, A., & Zisserman, A. (2013). Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps. arXiv Preprint arXiv:1312.6034, 1–8. Computer Vision and Pattern Recognition.
Code partially based on https://github.com/torch/nn/blob/master/Sequential.lua
Mnist example requires files from: https://github.com/e-lab/torch7-demos/tree/master/train-a-digit-classifier
Train the digit classifier and run the 'learnclassmodel.lua' with '-network logs/mnist.net'.