Investigation of the various layer outputs of the GoogLeNet Convolutional neural net trained on MIT Places dataset (http://places.csail.mit.edu/downloadCNN.html). code: https://github.com/graphific/DeepDreamVideo (first experiment on Fear and Loathing in Las Vegas at: https://www.youtube.com/watch?v=oyxSerkkP4o) Music in the video is also generated by a deep neural net. More songs & explanation a
![No more Dogs! Deep Dreaming on MIT Places CNN - YouTube](https://cdn-ak-scissors.b.st-hatena.com/image/square/155bf372e6a9e3a93a1e1f8abd6c8de0d79ecf69/height=288;version=1;width=512/https%3A%2F%2Fi.ytimg.com%2Fvi%2F6IgbMiEaFRY%2Fhqdefault.jpg)