/reinforcement_learning

MDP and Q-Learning algorithms developed during a project at UNSW in 2011

Primary LanguageC++GNU General Public License v2.0GPL-2.0

Reinforcement learning

MDP and Q-Learning algorithms developed during a project at UNSW in 2011.

Documentation

More documentation can be found in the PDF (Documentation.pdf) and in the html files in the doc folder.

Installation

mkdir build

cd build

cmake ..

make

Execution

To execute the test on the grid world, run

./demoGridWorldDemo --value-iteration | --poliy-iteration | --q-learning

Note

The source code is mainly contained in the header files at the moment, partly contaning several classes per header file. This is partly because the use of templates limits the possibilities to put code in cpp files, and partly because I find it easier to work with.