2023/08/17 by Vaidotas Šimkus, Michael U. Gutmann, Simkus, Vaidotas +1 · 1 citation
Computer Science · #62D10 #Bayesian Methods and Mixture Models #FOS: Computer and information sciences #G.3 #Gaussian Processes and Bayesian Inference #Generative Adversarial Networks and Image Synthesis #Machine Learning (cs.LG) #Machine Learning (stat.ML)
paper · pdf · doi:10.48550/arxiv.2308.09078
openalex publication_date 2023/08/17 · openalex created_date 2023/08/22 · openalex updated_date 2026/08/01
Conditional sampling of variational autoencoders (VAEs) is needed in various applications, such as missing data imputation, but is computationally intractable. A principled choice for asymptotically exact conditional sampling is Metropolis-within-Gibbs (MWG). However, we observe that the tendency of VAEs to learn a structured latent space, a commonly desired property, can cause the MWG sampler to get "stuck" far from the target distribution. This paper mitigates the limitations of MWG: we systematically outline the pitfalls in the context of VAEs, propose two original methods that address these pitfalls, and demonstrate an improved performance of the proposed methods on a set of sampling tasks.