Piecewise Linear Parametrization of Policies: Towards Interpretable Deep Reinforcement Learning.
Maxime WabarthaJoelle PineauPublished in: ICLR (2024)
Keyphrases
- piecewise linear
- reinforcement learning
- optimal policy
- dynamic programming
- policy search
- control policies
- markov decision process
- partially observable markov decision processes
- markov decision processes
- state space
- fitted q iteration
- reward function
- control policy
- chaotic map
- function approximation
- markov decision problems
- finite sets
- model free
- policy gradient methods
- reinforcement learning algorithms
- principal curves
- regression algorithm
- learning algorithm
- temporal difference
- hyperplane