bacow/easy-rl
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

update Sarsa

Browse Source
This commit is contained in:
JohnJim0816
2021-03-12 17:19:04 +08:00
parent 1fdcfbbd66
commit f1394feb65
10 changed files with 147 additions and 357 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
codes/Sarsa/README.md Normal file
View File

@@ -0,0 +1,19 @@
# Sarsa
## 使用说明
运行```main.py```即可
## 环境说明
见[环境说明](https://github.com/JohnJim0816/reinforcement-learning-tutorials/blob/master/env_info.md)中的The Racetrack
## 算法伪代码
![sarsa_algo](assets/sarsa_algo.png)
## 其他说明
### 与Q-learning区别
算法上区别很小，只在更新公式上，但Q-learning是Off-policy，而Sarsa是On-policy，可参考[知乎：强化学习中sarsa算法是不是比q-learning算法收敛速度更慢？](https://www.zhihu.com/question/268461866)