Joint Importance Sampling for Variational Inference

Jack Klys, Jesse Bettencourt, David Duvenaud

Feb 12, 2018 (modified: Feb 12, 2018) ICLR 2018 Workshop Submission readers: everyone
  • Abstract: We consider methods of variance reduction in Monte Carlo estimators which arise from importance sampling, with application to variational inference. We show that learning dependencies between samples, while preserving their marginal distributions outperforms sampling techniques which assume independence among samples in some settings.