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
@hardmaru When all you know are ten balanced categories, 10% of everything and anything looks like a frog.
-
-
@fchollet that's true if we look at argmax of softmax output, but was playing around with maximizing the inferred probability as fit func -
New conversation -
-
-
@fchollet it seems it is too easy for an algo to generate a random image to score highly on a convnet classifier. Just 4-5 gen to score 99%Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.