Experiments in running Deep Learning networks in reverse.
Most linear and non-linear operations used in Deep Learning networks are reversible. This project does some experiments to demonstrate this, and shows that running a DL network in reverse is a form of explanation.