PROJECTS

Snake AI

Artificial intelligence that plays snake.
This is not an artifical intelligence that learns how to play it but, given certain data, it finds the best way to achieve the objectives which are only 3 for this game: Eat the apples and avoid hitting the wall or himself.
To implementi it, I used python, more precisely pygame for the game and tensorflow for the AI.
This AI is a type of artificial intelligence that uses a preadded model. Is commonly called "model-based artificial intelligence" or "neural network-based artificial intelligence". This approach is part of supervised learning, where a model is trained on a dataset labeled and then used to make predictions about new data without further adaptation during use.
For "training" the AI use both fruit and snake position and their distance so the AI can calculate the best path to achieve the apple without hit something (By now the record of apple eaten is of 53 apples :) ).
Link to download