Deep neural networks are easily fooled: High confidence predictions for unrecognizable images.

Anh Mai Nguyen, Jason Yosinski, Jeff Clune

Browse the full CVPR paper archive.