A method to undestand how image classifiers work by finding which features activate individual neurons.
