January 2019
Intermediate to advanced
386 pages
11h 13m
English
book
Python Deep Learning - Second Edition
by Ivan Vasilev, Daniel Slater, Gianmario Spacagna, Peter Roelants, Valentino Zocca
Content preview from Python Deep Learning - Second Edition
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
Start your free trial
Control with Q-learning
Q-learning is an off-policy TD control method. It was developed by Watkins, C.J.C.H. (http://www.cs.rhul.ac.uk/~chrisw/new_thesis.pdf) in 1989 (in 2019, Q-learning is able to legally run for a seat in the United States Senate). With some improvements, it is one of the most popular RL algorithms in use today. As with Sarsa and MC, we have to estimate the action-value function. Q-learning is defined as follows:
Although it resembles the definition of Sarsa, it has one substantial difference: tt is an off-policy method, which means that we have two distinct policies:
- Behavior policy: The agent uses this to actually navigate ...