Denoising Criterion for Variational Auto-Encoding Framework
This is an incremental improvement for researchers in generative models, addressing training intractability in noisy setups.
The paper tackles the problem of improving variational autoencoders by injecting noise both at the input and hidden layers, proposing a modified variational lower bound for tractable training. The result shows that the denoising variational autoencoder yields better average log-likelihood than VAE and importance weighted autoencoder on MNIST and Frey Face datasets.
Denoising autoencoders (DAE) are trained to reconstruct their clean inputs with noise injected at the input level, while variational autoencoders (VAE) are trained with noise injected in their stochastic hidden layer, with a regularizer that encourages this noise injection. In this paper, we show that injecting noise both in input and in the stochastic hidden layer can be advantageous and we propose a modified variational lower bound as an improved objective function in this setup. When input is corrupted, then the standard VAE lower bound involves marginalizing the encoder conditional distribution over the input noise, which makes the training criterion intractable. Instead, we propose a modified training criterion which corresponds to a tractable bound when input is corrupted. Experimentally, we find that the proposed denoising variational autoencoder (DVAE) yields better average log-likelihood than the VAE and the importance weighted autoencoder on the MNIST and Frey Face datasets.