Is this CIFAR10's Frog? Reverse Engineering CIFAR10 convnet for Image Generation http://blog.otoro.net/2015/06/21/reverse-engineering-cifar-10-convnet/ …pic.twitter.com/cfwzE58dHJ
-
@fchollet I'm thinking of a way to get a nnet to generate random images that look similar to real images. Perhaps train a classifier first
@hardmaru intuitively, i think you'll need much more than cifar10. Maybe look at the trained Caffe models for imagenet...