Skip to content

Latest commit

 

History

History
46 lines (25 loc) · 844 Bytes

File metadata and controls

46 lines (25 loc) · 844 Bytes

QLearning

Human-Level Control Through Deep Reinforcement Learning

Compiling

ALE_HOME=... CAFFE_HOME=.. ./configure

For example:

ALE_HOME=~/Workspace/ale_0_5 CAFFE_HOME=~/Workspace/caffe ./configure
cd build
make

Usage

./qcli learn
./qcli evaluate

Todo

Reference

Human-level control through deep reinforcement learning

Playing Atari with Deep Reinforcement Learning

http://caffe.berkeleyvision.org/

http://www.arcadelearningenvironment.org/

https://github.com/kuz/DeepMind-Atari-Deep-Q-Learner

https://github.com/muupan/dqn-in-the-caffe

http://www.infoq.com/cn/news/2014/10/deepmind/

GitHub project hpicgs/cmake-init is used to generate the cross-platform CMake files.

Author

Jiguanglizipao jiguanglizipao@163.com

Weijun Sun swj14@mails.tsinghua.edu.cn

Jason Lau i@dotkrnl.com