Auxiliary Deep Generative Models
This work addresses the problem of limited expressiveness in variational approximations for deep generative models, benefiting researchers in unsupervised and semi-supervised learning, though it appears incremental as it builds on existing models.
The paper tackled improving deep generative models by introducing auxiliary variables to enhance variational approximations, resulting in state-of-the-art performance in semi-supervised learning on MNIST, SVHN, and NORB datasets with faster convergence and better results.
Deep generative models parameterized by neural networks have recently achieved state-of-the-art performance in unsupervised and semi-supervised learning. We extend deep generative models with auxiliary variables which improves the variational approximation. The auxiliary variables leave the generative model unchanged but make the variational distribution more expressive. Inspired by the structure of the auxiliary variable we also propose a model with two stochastic layers and skip connections. Our findings suggest that more expressive and properly specified deep generative models converge faster with better results. We show state-of-the-art performance within semi-supervised learning on MNIST, SVHN and NORB datasets.