Integrating Behavior Cloning and Reinforcement Learning for Improved Performance in Dense and Sparse Reward Environments.
Vinicius G. GoecksGregory M. GremillionVernon J. LawhernJohn ValasekNicholas R. WaytowichPublished in: AAMAS (2020)
Keyphrases
- reinforcement learning
- multi agent environments
- robot behavior
- reinforcement learning algorithms
- eligibility traces
- state space
- autonomous robots
- real robot
- motion field estimation
- partially observable environments
- function approximation
- sparse data
- dense stereo
- transfer learning
- high dimensional
- reward function
- real world
- machine learning
- learning algorithm
- multi agent
- temporal difference
- supervised learning
- optimal control
- markov decision processes
- model free
- learning process
- long run
- behavior patterns
- sparse coding
- dense optical flow
- sparse representation
- dynamic environments