Reinforcement Learning agent for 2048 game
C++ implementation of Temporal Difference Learning for the 2048 game.
Based on paper: http://www.cs.put.poznan.pl/mszubert/pub/szubert2014cig.pdf
To compile program cd to project directory and write:
mkdir build; cd buid
cmake ..
make
Then to run program:
./learn <print_every> <learning_rate> <save_every>
print_every - after this many games print statistics
learning_rate - learning rate for algorithm
save_every - after this many games save model
Example usage:
./learn 100 0.01 1000