DRL-AI

This project is a CUDA implementation of the paper

Playing Atari with Deep Reinforcement Learning

Volodymyr Mnih, Koray Kavukcuoglu, David Silver, Alex Graves, Ioannis Antonoglou, Daan Wierstra, Martin Riedmiller (NIPS workshop 2013)

However we have not got the performance reported by the paper yet.

Authors

Yuan Gao & Yao Lu

Required Libraries

rl-glue, SDL, SDL-image, CUDA, cublas

Compile

cd ENV/src/dnq_gpu
make
cd ../..
make

Run

For example, to play game Breakout

./ale -player_agent deep_q_agent -display_screen true breakout.bin

About

Languages

Language:HTML 35.9%Language:C++ 21.7%Language:Shell 17.1%Language:Prolog 5.6%Language:Java 4.3%Language:TeX 4.0%Language:C 2.9%Language:Common Lisp 2.5%Language:Python 2.0%Language:MATLAB 1.8%Language:Makefile 1.2%Language:D 0.5%Language:Cuda 0.3%Language:Perl 0.1%Language:CSS 0.1%Language:Perl 6 0.0%