A Restart-based Rank-1 Evolution Strategy for Reinforcement Learning.
Zefeng ChenYuren ZhouXiaoyu HeSiyu JiangPublished in: IJCAI (2019)
Keyphrases
- evolution strategy
- reinforcement learning
- evolutionary algorithm
- cma es
- function approximation
- random walk
- genetic algorithm
- differential evolution
- genetic operators
- state space
- machine learning
- model free
- particle swarm optimization algorithm
- multi objective
- reinforcement learning algorithms
- optimization methods
- temporal difference
- global search
- optimization problems
- dynamic programming
- function approximators
- rank order
- candidate solutions
- highly ranked
- optimal control
- feature selection
- multi agent systems
- feasible solution
- learning problems
- markov decision processes
- computational complexity
- simulated annealing
- cost function