GEP-PG: Decoupling Exploration and Exploitation in Deep Reinforcement Learning Algorithms.
Cédric ColasOlivier SigaudPierre-Yves OudeyerPublished in: ICML (2018)
Keyphrases
- reinforcement learning algorithms
- reinforcement learning
- model free
- state space
- markov decision processes
- gene expression programming
- temporal difference
- function approximation
- eligibility traces
- reinforcement learning problems
- learning algorithm
- reinforcement learning methods
- action selection
- reward function
- policy search
- dynamic environments
- partially observable environments
- tabula rasa
- evolutionary computation
- reward shaping
- machine learning
- learning tasks
- genetic programming