The aim of the thesis is to design and implement an artificial intelligence (AI) for a computer game called Elementals based on evolutionary generation of neural networks. The Elementals game is a two player tactical game similar to the combat system in the game of Heroes of Might and Magic. Elementals is grid based with each tile having a distinct element. The Player controls a few elementals who gain different abilities by absorbing the elements present in the tiles. The objective of the game is to eliminate all opposing elementals.

The thesis work will evaluate the viability of generated neural networks for turn based computer games as well as their performance and the difficulty of their creation via evolution.

The whole thesis will consists of two parts - a textual part and a practical implementation of a functional module of the game AI. The theoretical part will present an introduction of underlying technologies, an overview of existing related works, an analysis and implementation details of the developed module together with an evaluation of its efficiency. The practical part will consist in implementation of the AI for the Elementals game.