Practical_RL/week1 at master · ghellstern/Practical_RL · GitHub

Name		Name	Last commit message	Last commit date
parent directory ..
README.md		README.md
breakout.py		breakout.py
crossentropy_method.ipynb		crossentropy_method.ipynb

README.md

Materials:

Lecture slides
Lecture and seminar videos (russian) - lecture, seminar only covering crossentropy method
[main] Lecture by J Schulman with crossentropy method explained (english) - https://www.youtube.com/watch?v=aUrX-rP_ss4&list=PLCTc_C7itk-GaAMxmlChrkPnGKtjz8hv1
All monte-carlo methods - https://webdocs.cs.ualberta.ca/~sutton/book/ebook/node50.html
Article about CEM in general - https://people.smp.uq.edu.au/DirkKroese/ps/eormsCE.pdf
Article about CEM for optimization - https://people.smp.uq.edu.au/DirkKroese/ps/CEopt.pdf
Article about CEM in reinforcement learning - http://www.aaai.org/Papers/ICML/2003/ICML03-068.pdf

Homework description

Just follow the notebook
During the CartPole-v0 section (and similar envs), a window will pop up, displaying some game state. The window won't respond to direct input and is instead changes each time you call env.render(). Don't force-close this process, just ignore it until you complete the notebook.
important the current newest version of gym force-stops environment in 200 steps even if you don't use env.monitor.
- This may ruin CEM on MountainCar. To avoid this, use gym.make("MountainCar-v0").env