Knowledge Compilation for Constrained Combinatorial Action Spaces in Reinforcement Learning.
Jiajing LingMoritz Lukas SchulerAkshat KumarPradeep VarakanthamPublished in: AAMAS (2023)
Keyphrases
- action space
- knowledge compilation
- reinforcement learning
- state space
- product configuration
- state and action spaces
- markov decision processes
- continuous state
- real valued
- action selection
- stochastic processes
- prime implicates
- state action
- dynamic programming
- constraint satisfaction
- function approximators
- continuous state spaces
- normal form
- model free
- machine learning
- skill learning
- decomposable negation normal form
- optimal policy
- decision support
- markov decision process
- single agent
- heuristic search
- continuous action
- optimal control
- process planning
- markov decision problems
- constraint satisfaction problems
- dynamic environments
- knowledge based systems
- sufficient conditions
- information retrieval