2020/03/19 by Luis A. Pérez Rey, Rey, Luis A. Pérez
Computer Science · Mathematics · Physics and Astronomy · #FOS: Computer and information sciences #Generative Adversarial Networks and Image Synthesis #Image Processing and 3D Reconstruction #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Model Reduction and Neural Networks #cs.LG #stat.ML
paper · pdf · doi:10.48550/arxiv.2003.08996
arxiv created 2020/03/19 · openalex publication_date 2020/03/19 · arxiv updated 2020/03/23 · openalex created_date 2020/03/27 · openalex updated_date 2026/07/28
A disentangled representation of a data set should be capable of recovering the underlying factors that generated it. One question that arises is whether using Euclidean space for latent variable models can produce a disentangled representation when the underlying generating factors have a certain geometrical structure. Take for example the images of a car seen from different angles. The angle has a periodic structure but a 1-dimensional representation would fail to capture this topology. How can we address this problem? The submissions presented for the first stage of the NeurIPS2019 Disentanglement Challenge consist of a Diffusion Variational Autoencoder (ΔVAE) with a hyperspherical latent space which can, for example, recover periodic true factors. The training of the ΔVAE is enhanced by incorporating a modified version of the Evidence Lower Bound (ELBO) for tailoring the encoding capacity of the posterior approximate.