On the Expressivity of Objective-Specification Formalisms in Reinforcement Learning.
Rohan SubramaniMarcus WilliamsMax HeitmannHalfdan HolmCharlie GriffinJoar Max Viktor SkalsePublished in: ICLR (2024)
Keyphrases
- reinforcement learning
- computational properties
- high level
- machine learning
- knowledge representation
- logic programming
- specification language
- formal specification
- function approximation
- logic programs
- multiple objectives
- markov decision processes
- state space
- neural network
- learning process
- multi agent
- action selection
- reinforcement learning algorithms
- artificial intelligence
- learning algorithm
- formal verification