Reinforcement learning on a tiny problem: returns, values and Q-learning

Before eight servos, a 4 by 4 floor with two holes. Discounted returns, value functions and Q-learning, worked out by hand and then in 50 lines of Python.

50 min

Táto stránka ešte nie je preložená, preto je zobrazená v angličtine.