MT-Opt: Continuous Multi-Task Robotic Reinforcement Learning at Scale.
Dmitry KalashnikovJacob VarleyYevgen ChebotarBenjamin SwansonRico JonschkowskiChelsea FinnSergey LevineKarol HausmanPublished in: CoRR (2021)
Keyphrases
- multi task
- reinforcement learning
- transfer learning
- multi task learning
- learning problems
- learning tasks
- multitask learning
- multiple tasks
- learning algorithm
- feature selection
- multi class
- sparse learning
- gaussian processes
- supervised learning
- labeled data
- data mining
- markov random field
- pairwise
- decision trees
- machine learning