Poster
Learning Latent Subspaces in Variational Autoencoders
Jack Klys · Jake Snell · Richard Zemel

Thu Dec 6th 10:45 AM -- 12:45 PM @ Room 517 AB #147

Variational autoencoders (VAEs) are widely used deep generative models capable of learning unsupervised latent representations of data. Such representations are often difficult to interpret or control. We consider the problem of unsupervised learning of features correlated to specific labels in a dataset. We propose a VAE-based generative model which we show is capable of extracting features correlated to binary labels in the data and structuring it in a latent subspace which is easy to interpret. Our model, the Conditional Subspace VAE (CSVAE), uses mutual information minimization to learn a low-dimensional latent subspace associated with each label that can easily be inspected and independently manipulated. We demonstrate the utility of the learned representations for attribute manipulation tasks on both the Toronto Face and CelebA datasets.

Author Information

Jack Klys (University of Toronto)
Jake Snell (University of Toronto, Vector Institute)
Richard Zemel (Vector Institute/University of Toronto)

More from the Same Authors