Undirected Graphical Models as Approximate Posteriors
This work addresses the challenge of poor generative performance in VAEs due to posterior mismatch, offering a novel training method for researchers in machine learning and generative modeling.
The authors tackled the problem of improving variational autoencoders (VAEs) by using undirected graphical models as approximate posteriors, demonstrating that this approach outperforms previous directed models in discrete VAEs.
The representation of the approximate posterior is a critical aspect of effective variational autoencoders (VAEs). Poor choices for the approximate posterior have a detrimental impact on the generative performance of VAEs due to the mismatch with the true posterior. We extend the class of posterior models that may be learned by using undirected graphical models. We develop an efficient method to train undirected approximate posteriors by showing that the gradient of the training objective with respect to the parameters of the undirected posterior can be computed by backpropagation through Markov chain Monte Carlo updates. We apply these gradient estimators for training discrete VAEs with Boltzmann machines as approximate posteriors and demonstrate that undirected models outperform previous results obtained using directed graphical models. Our implementation is available at https://github.com/QuadrantAI/dvaess .