We propose to learn a hierarchical prior in the context of variational autoencoders to avoid the over-regularisation resulting from a standard normal prior distribution. To incentivise an informative latent representation of the data, we formulate the learning problem as a constrained optimisation problem by extending the Taming VAEs framework to two-level hierarchical models. We introduce a graph-based interpolation method, which shows that the topology of the learned latent representation corresponds to the topology of the data manifold---and present several examples, where desired properties of latent representation such as smoothness and simple explanatory factors are learned by the prior.
@article{arxiv.1905.04982,
title = {Learning Hierarchical Priors in VAEs},
author = {Alexej Klushyn and Nutan Chen and Richard Kurle and Botond Cseke and Patrick van der Smagt},
journal= {arXiv preprint arXiv:1905.04982},
year = {2019}
}