PROCEDURAL CONTENT GENERATION
The
Forza
franchise is an excellent example of how deep learning using ANNs (Artificial Neural Networks) can be used to adapt to player behavior, providing better replay value and a more enjoyable gameplay experience.
In the case of the AI racing cars in
Forza Motorsport
, the process used is a form of ‘reinforcement’ learning by analyzing past player driving performance and trying out strategies to navigate obstacles like corners to find an optimal approach.
Another exciting aspect of machine learning in videogames relies on using PCG (procedural content generation). One of the earliest examples of this was
Rogue
(1980). The game’s algorithm worked by generating a dungeon room into which the player would spawn. It would iterate at random through more rooms, determining if each was accessible by its neighbor, then placing stairs in the last room, so players could proceed to the next level.