2018 (modified: 11 Nov 2022)ICML 2018Readers: Everyone
Abstract:We define and address the problem of unsupervised learning of disentangled representations on data generated from independent factors of variation. We propose FactorVAE, a method that disentangles ...