Learning the Stein Discrepancy for Training and Evaluating Energy-Based Models without Sampling.
Will GrathwohlKuan-Chieh WangJörn-Henrik JacobsenDavid DuvenaudRichard S. ZemelPublished in: ICML (2020)
Keyphrases
- prior knowledge
- structured prediction
- supervised learning
- online learning
- feedforward neural networks
- learning speed
- learning algorithm
- learning process
- training process
- learning models
- learned models
- learning systems
- deep architectures
- inductive inference
- probabilistic model
- recurrent networks
- accurate models
- complex systems
- learning problems
- training examples
- unsupervised learning
- hidden markov models
- learning environment
- reinforcement learning