On Multi-objective Policy Optimization as a Tool for Reinforcement Learning.
Abbas AbdolmalekiSandy H. HuangGiulia VezzaniBobak ShahriariJost Tobias SpringenbergShruti MishraDhruva TBArunkumar ByravanKonstantinos BousmalisAndrás GyörgyCsaba SzepesváriRaia HadsellNicolas HeessMartin A. RiedmillerPublished in: CoRR (2021)
Keyphrases
- multi objective
- optimization algorithm
- reinforcement learning
- optimal policy
- multiple objectives
- evolutionary algorithm
- evolutionary optimization
- conflicting objectives
- optimum design
- multiobjective optimization
- engineering design problems
- multi objective optimization
- state space
- policy search
- optimization problems
- action selection
- multi objective evolutionary algorithms
- multi agent
- markov decision process
- policy iteration
- genetic algorithm
- function approximation
- global optimization
- markov decision processes
- nsga ii
- control policy
- particle swarm optimization
- approximate dynamic programming
- markov decision problems
- learning algorithm
- objective function
- function approximators
- dynamic programming
- fitness function
- state and action spaces
- pareto optimal
- machine learning
- multi objective evolutionary
- partially observable environments
- continuous state
- engineering problems
- estimation of distribution algorithms
- reward function
- long run
- optimal control
- combinatorial optimization