Interpretable Policy Extraction with Neuro-Symbolic Reinforcement Learning.
Rajdeep DuttaQincheng WangAnkur SinghDhruv KumarjigudaXiaoli LiSenthilnath JayaveluPublished in: ICASSP (2024)
Keyphrases
- reinforcement learning
- optimal policy
- policy search
- action selection
- markov decision process
- actor critic
- policy gradient
- markov decision processes
- reinforcement learning algorithms
- temporal difference
- state and action spaces
- reinforcement learning problems
- partially observable
- partially observable environments
- control policies
- policy iteration
- function approximation
- state space
- approximate dynamic programming
- policy evaluation
- markov decision problems
- rl algorithms
- reward function
- artificial neural networks
- automatic extraction
- continuous state spaces
- information extraction
- function approximators
- model free
- decision problems
- action space
- optimal control
- agent learns
- learning algorithm
- partially observable markov decision processes
- average reward
- neural network
- reinforcement learning methods
- supervised learning
- continuous state
- high level
- state action
- dynamic programming
- temporal difference learning
- sufficient conditions
- control problems
- neuro fuzzy
- infinite horizon
- symbolic representation
- finite state