GEP-PG: Decoupling Exploration and Exploitation in Deep Reinforcement Learning Algorithms.
Cédric ColasOlivier SigaudPierre-Yves OudeyerPublished in: CoRR (2018)
Keyphrases
- reinforcement learning algorithms
- reinforcement learning
- state space
- model free
- gene expression programming
- markov decision processes
- temporal difference
- eligibility traces
- reinforcement learning problems
- learning algorithm
- reinforcement learning methods
- function approximation
- dynamic environments
- partially observable environments
- genetic programming
- policy search
- action selection
- reward function
- evolutionary computation
- feature selection
- optimal policy
- function approximators
- artificial neural networks
- multi agent
- multiagent reinforcement learning
- machine learning