Data Driven Reward Initialization for Preference based Reinforcement Learning.
Mudit VermaSubbarao KambhampatiPublished in: CoRR (2023)
Keyphrases
- reinforcement learning
- data driven
- function approximation
- learning algorithm
- state space
- eligibility traces
- model free
- reinforcement learning algorithms
- markov decision processes
- reward function
- k means
- temporal difference
- policy search
- initial conditions
- total reward
- reward shaping
- machine learning
- learning agent
- average reward
- policy iteration
- partially observable environments
- partially observable
- learning capabilities
- learning problems
- user preferences
- learning process
- multi agent
- action selection
- optimal control
- reinforcement learning methods
- policy gradient
- multi agent reinforcement learning
- transfer learning
- genetic algorithm