Neural-Network Reinforcement Learning Library
##Setup
To run demonstrations with the NeuRL library, Rl-Glue and the 2009 AI Competition software must be installed. These are included in the 'external' directory for convenience. These files can be installed from the project home directory with:
$ sudo ./setup.sh
##Building
The C++ source code for the project can be built with CMake. To do this from the home directory:
mkdir build
cd build
cmake ..
make
##Running Demo Experiments
Demonstration experiments are in the 'experiments' directory. They can be run with python executables, eg:
$ python run_mario_exp.py
##Roadmap
- C++ Implementation of Neural Q-Learning
- Customizable Experience Weighting
- GPU-Acceleration with cuDNN
- Atari ALE Experiments