31. Autoencoders and variational autoencoders

Compress data into latent variables with deterministic autoencoders and probabilistic variational autoencoders. Use reconstruction, regularized latent spaces, sampling, and interpolation to study learned structure.