Neural Network Design
Start Date: September 2015
End Date: December 2015 Description: Developed a neural network to play and improve at the game 'Minesweeper'. Continually improving code for more advanced learning algorithm.
Link to GitHub Repo
The neural net model consists of 'tanks' that initially have a randomized direction andorientation. These tanks are rewarded once they come into contact with a green square. The tank that came in to contact with the most green squares is deemed the fittest of that generation. The fittest, then determines how the average tank moves the next iteration. If you download my code and run the executable, by pressin 'F'' on the keyboard you can speed up the generation interations and see the best fit and average fitness values. Although this was just for a course in Neural Networks Engineering, I will without a doubt use these principles of machine learning in my future projects.