2
android
1
biopython
3
bootstrap
1
c-plus-plus
2
c-sharp
6
css
2
cv2
2
firebase
1
flex-lexical-analysis
6
fontawesome
2
game
2
google-maps
3
hobby
6
html
1
jasypt
8
java
1
java-spark
6
javascript
3
jquery
1
jstl
1
laravel
3
matplotlib
1
mockaroo
2
multithreading
3
mysql
1
neat-algorithm
3
netlify
1
networkx
1
neural-networks
6
numpy
2
pandas
1
php
2
pillow
2
postgresql
9
python
3
react
1
reinforcement-learning
1
scikit-image
2
scikit-learn
1
scipy
2
servlets
7
sql
2
sqlite
3
tailwindcss
2
tomcat
19
university
3
visual-studio
6
web-app
2
website
Thesis
Involved in this project: Dionysios Sotiropoulos
On this project, we study applications of the NEAT algorithm in deterministic and non-deterministic game environments. First, we look at an overview of the NEAT algorithm, how it works, design principles and the challenges that come with implementation. Next, we introduce a custom two-dimensional game in Python for two players: blue and red. We lay down the basic rules and structure, in order to create an environment suitable for neuroevolution. Finally, we study five training cases, where the blue and red player are given several tasks that must be achieved through the evolution of neural networks.
Tic-Tac-Toe
A simple Tic-Tac-Toe game implementation in C# using window based components. The user is greeted with the game setup form to select the blue or red player. Then, the game starts off in a five by five grid of clickable elements, similar to the traditional Tic-Tac-Toe setup. The user is tasked to beat a naive computer player that is configured to make random moves by selecting empty spots on the game board. As the game progresses, we check whether our win conditions are satisfied by checking diagonal left, diagonal right, horizontally and vertically. Depending on each player's moves, the game can end in a win or draw state.