On Incorporating Inductive Biases into VAEsDownload PDF

29 Sept 2021, 00:32 (edited 14 Feb 2022)ICLR 2022 PosterReaders: Everyone
  • Keywords: VAEs, Variational autoencoders, Variational auto-encoders, Representation learning, Inductive biases
  • Abstract: We explain why directly changing the prior can be a surprisingly ineffective mechanism for incorporating inductive biases into variational auto-encoders (VAEs), and introduce a simple and effective alternative approach: Intermediary Latent Space VAEs (InteL-VAEs). InteL-VAEs use an intermediary set of latent variables to control the stochasticity of the encoding process, before mapping these in turn to the latent representation using a parametric function that encapsulates our desired inductive bias(es). This allows us to impose properties like sparsity or clustering on learned representations, and incorporate human knowledge into the generative model. Whereas changing the prior only indirectly encourages behavior through regularizing the encoder, InteL-VAEs are able to directly enforce desired characteristics. Moreover, they bypass the computation and encoder design issues caused by non-Gaussian priors, while allowing for additional flexibility through training of the parametric mapping function. We show that these advantages, in turn, lead to both better generative models and better representations being learned.
  • One-sentence Summary: A flexible and effective framework for adding inductive biases to VAEs that corrects the pathologies of previous approaches and leads to improved representations and generative models.
  • Supplementary Material: zip
20 Replies