no code implementations • 15 Feb 2018 • Lise Aubin, Mehdi Khamassi, Benoît Girard
The Dyna reinforcement learning algorithms use off-line replays to improve learning.
Hippocampus Q-Learning +2