OMPO: A Unified Framework for RL under Policy and Dynamics Shifts.
Yu LuoTianying JiFuchun SunJianwei ZhangHuazhe XuXianyuan ZhanPublished in: CoRR (2024)
Keyphrases
- optimal policy
- reinforcement learning
- markov decision process
- action selection
- policy search
- action space
- markov decision processes
- control policy
- dynamical systems
- decision problems
- policy gradient
- policy iteration
- actor critic
- state space
- partially observable domains
- asymptotically optimal
- policy evaluation
- rl algorithms
- state and action spaces
- model free
- least squares
- policy makers
- state dependent
- finite state
- reward function
- dynamic model
- markov decision problems
- sequential decision making
- function approximation
- genetic algorithm
- multi agent
- inverse reinforcement learning
- learning process
- approximate dynamic programming
- markov chain
- control policies
- learning agents
- function approximators
- learning classifier systems
- temporal difference
- partially observable markov decision processes
- machine learning
- partially observable