/Battleships

Can a neural network sink your battleship?

Primary LanguagePython

Battleships

==============
Battleships is an experiment with neural networks, trying to get the neural networks to guess the location of battleships, similarly to the board game Battleship.

============== Requirements ==============

  1. Python 3.3 or higher
  2. Google's TensorFlow

============== Running ==============
Steps for running Battleships are as follows

  1. Create a training set with usingtf/traininggenerator.py.
  2. Neural networks are then trained using Battleship.py. In Battleship.py, the function startEvolution() initializes the neural networks and saves them to NNetmain. The function continueEvolution will then continue the neural networks' evolution.
  3. Visualizer.py allows for visual tests of a particular neural network. It takes as input a neural network and outputs its score for a particular round.

============== Credits ==============
Development of this Python suite of codes was done by Daniel Neshyba-Rowe and is freely available. Please share any suggestions for improvement.