ShipzAI is a reinforcment learning for the classic game Battleships.
ShipzAI uses the OpenAI Gym toolkit and Stable Baselines
for the implementation.
To install the dependencies typ:
pip install gym numpy tensorflow==1.13.2 stable-baselines
After installtion, ShipzAI can be run like:
python Play.py
Model can be trained with e.g.:
python TrainACKTR.py
The progress of the training can be observed with Tensorboard:
tensorboard --port 6004 --logdir ./logs/progress_tensorboard/
If a game is finished with a negativ reward/score an invalid action (shooting same field multiple times)
was executed.
Philipp Doll MatrNr: 700911
Christian Leich MatNr: 699570
Max van Aerssen MatrNr: 699795
Group-ID: 11