February 2020
Intermediate to advanced
432 pages
10h 50m
English
The script runs 100,000 episodes in less than one minute. Then, we evaluate the agent with 100 episodes and obtain these average values:
In front of the brute-force approach (which you can find in the Jupyter notebook), you obtain routes that vary from hundreds to thousands of steps and more than 1,000 penalties (remember that a penalty of 1 was given for each illegal action incurred).
Read now
Unlock full access