Login / Signup

One-Line-of-Code Data Mollification Improves Optimization of Likelihood-based Generative Models.

Ba-Hien TranGiulio FranzesePietro MichiardiMaurizio Filippone
Published in: CoRR (2023)
Keyphrases
  • generative model
  • data sets
  • training data
  • probabilistic model
  • prior knowledge
  • learning algorithm
  • object detection
  • maximum likelihood
  • mixture model
  • multi label
  • hidden variables