Shaping in Reinforcement Learning by Changing the Physics of the Problem.
Jette RandløvPublished in: ICML (2000)
Keyphrases
- reinforcement learning
- reward shaping
- reinforcement learning algorithms
- function approximation
- state space
- markov decision processes
- temporal difference
- computer science
- machine learning
- model free
- multi agent
- artificial intelligence
- information retrieval
- neural network
- policy search
- markov decision problems
- three dimensional
- optimal policy
- case study
- genetic algorithm
- learning environment
- complex domains
- control problems
- policy iteration
- function approximators
- temporal difference learning
- supervised learning
- hidden markov models