On Double Descent in Reinforcement Learning with LSTD and Random Features.
David BrellmannEloïse BerthierDavid FilliatGoran FrehsePublished in: ICLR (2024)
Keyphrases
- reinforcement learning
- temporal difference
- function approximation
- state space
- temporal difference learning
- model free
- least squares
- control problems
- reinforcement learning algorithms
- multi agent
- policy evaluation
- optimal control
- optimal policy
- evaluation function
- supervised learning
- markov decision processes
- regression model
- markov chain
- policy iteration
- function approximators