cartpole-qlearning-master
所属分类：其他
开发工具：Python
文件大小：3KB
下载次数：1
上传日期：2021-04-29 10:26:42
上 传 者：
emrayo
说明： 深度强化学习DQN在倒立摆上的实现，python代码，使用torch，gym库
(Implementation of deep reinforcement learning dqn on inverted pendulum)
文件列表:
LICENSE.md (1072, 2019-12-19)
cartpole_qlearning (0, 2019-12-19)
cartpole_qlearning\__init__.py (0, 2019-12-19)
cartpole_qlearning\agent.py (841, 2019-12-19)
cartpole_qlearning\main.py (2524, 2019-12-19)
requirements.txt (144, 2019-12-19)
# cartpole-qlearning
The result of some playing around with [Gym](https://github.com/openai/gym "OpenAI Gym") to train an agent to balance an inverted pendulum using [Q-Learning](https://en.wikipedia.org/wiki/Q-learning "Wikipedia").
The current code manages to solve the problem by only using the angle and the angular velocity of the pole, completely ignoring the linear position and velocity of the cart (to reduce dimensionality for faster convergence). I'm sure it could be tweaked even further to improve the results.
近期下载者：
相关文件：
收藏者：