Learning in Variational Autoencoders with Kullback-Leibler and Renyi Integral Bounds
This addresses a specific technical issue for researchers and practitioners using VAEs in image generation, though it appears incremental as it builds on existing variational inference methods.
The authors tackled the problem of numerical instability in training Variational Autoencoders (VAEs) on high-contrast continuous datasets like handwritten digits, proposing novel bounds based on Kullback-Leibler and Rényi divergences that enable stable training without adding extra noise.
In this paper we propose two novel bounds for the log-likelihood based on Kullback-Leibler and the Rényi divergences, which can be used for variational inference and in particular for the training of Variational AutoEncoders. Our proposal is motivated by the difficulties encountered in training VAEs on continuous datasets with high contrast images, such as those with handwritten digits and characters, where numerical issues often appear unless noise is added, either to the dataset during training or to the generative model given by the decoder. The new bounds we propose, which are obtained from the maximization of the likelihood of an interval for the observations, allow numerically stable training procedures without the necessity of adding any extra source of noise to the data.