Learning Multiple Behaviors from Unlabeled Demonstrations in a Latent Controller Space.
Javier AlmingolLuis MontesanoManuel LopesPublished in: ICML (3) (2013)
Keyphrases
- active learning
- learning algorithm
- learning process
- knowledge acquisition
- supervised learning
- prior knowledge
- learning tasks
- space time
- labeled data
- unsupervised learning
- search space
- multiple tasks
- human behavior
- neural nets
- control algorithm
- labeled and unlabeled data
- control strategy
- learning problems
- real time
- delayed feedback
- learning systems
- unlabeled data
- online learning
- low dimensional
- reinforcement learning
- training data
- decision trees
- data sets