Deep Autoencoders: From Understanding to Generalization Guarantees - 42Papers