Comment by CodesInChaos

The way you'd do compression using a NN, is using the NN to predict the probability of the next symbol, and feeding that into an arithmetic coder to produce a compressed representation. This process is lossless, and better prediction quality directly translates into better compression.