Improving Exploration in Evolution Strategies for Deep Reinforcement Learning via a Population of Novelty-Seeking Agents.
Edoardo ContiVashisht MadhavanFelipe Petroski SuchJoel LehmanKenneth O. StanleyJeff ClunePublished in: NeurIPS (2018)
Keyphrases
- evolution strategy
- reinforcement learning
- action selection
- multi agent
- evolutionary algorithm
- learning agents
- multi agent systems
- multiagent systems
- evolutionary programming
- single agent
- differential evolution
- learning agent
- multi agent environments
- exploration strategy
- multi agent reinforcement learning
- intelligent agents
- entire population
- agent receives
- genetic algorithm
- multiple agents
- autonomous agents
- numerical optimization
- cooperative
- multiagent learning
- active exploration
- population dynamics
- software agents
- robocup soccer
- temporal difference
- learning capabilities
- state space
- mutation operator
- function approximation
- reinforcement learning algorithms
- optimization methods
- learning algorithm
- neural network
- autonomous learning
- multi objective
- particle swarm optimization
- optimization algorithm
- markov decision processes
- global optimization
- search strategies