A Demonstration of Issues with Value-Based Multiobjective Reinforcement Learning Under Stochastic State Transitions.
Peter VamplewCameron FoaleRichard DazeleyPublished in: CoRR (2020)
Keyphrases
- state transitions
- multi objective
- state transition
- reinforcement learning
- state space
- evolutionary algorithm
- direct policy search
- state action
- input output
- multiobjective optimization
- stochastic approximation
- function approximation
- machine learning
- optimization algorithm
- multiple objectives
- multi objective optimization
- reinforcement learning algorithms
- particle swarm optimization
- markov chain
- conflicting objectives
- black box
- markov decision processes
- monte carlo
- multiobjective evolutionary algorithm
- dynamic programming
- model free
- temporal difference
- bi objective
- objective function
- differential evolution
- database
- optimal policy
- genetic algorithm
- hidden markov models
- learning algorithm
- multiobjective evolutionary algorithms
- activity recognition
- markov decision process
- pareto optimal
- finite state machines
- probability distribution
- relational databases
- databases