Assessing Policy, Loss and Planning Combinations in Reinforcement Learning Using a New Modular Architecture.
Tiago Gaspar OliveiraArlindo L. OliveiraPublished in: EPIA (2022)
Keyphrases
- modular architecture
- reinforcement learning
- action selection
- reinforcement learning problems
- optimal policy
- partially observable
- partially observable markov decision processes
- reinforcement learning algorithms
- markov decision problems
- policy search
- partially observable environments
- function approximation
- reinforcement learning methods
- function approximators
- reward function
- action space
- temporal difference
- control policies
- partially observable domains
- state and action spaces
- policy iteration
- markov decision process
- conceptual framework
- planning problems
- state space
- control policy
- markov decision processes
- loosely coupled
- deterministic domains
- stochastic domains
- state action
- actor critic
- optimal control
- partial observability
- policy evaluation
- learning algorithm
- policy gradient
- temporal difference learning
- infinite horizon
- model free
- machine learning
- continuous state
- decision problems
- average reward
- planning domains
- dynamic programming
- domain knowledge