Combining manual feedback with subsequent MDP reward signals for reinforcement learning.
W. Bradley KnoxPeter StonePublished in: AAMAS (2010)
Keyphrases
- reinforcement learning
- markov decision processes
- reward function
- optimal policy
- state space
- markov decision process
- function approximation
- average reward
- action sets
- reinforcement learning algorithms
- signal processing
- semi automatic
- temporal difference
- partially observable
- total reward
- multi agent
- machine learning
- sensory inputs
- learning algorithm
- finite state
- policy iteration
- dynamic programming
- long run
- optimal control
- state action
- eligibility traces
- learning process
- partially observable environments
- bayesian reinforcement learning
- learning capabilities
- error prone
- planning under uncertainty