In machine learning there are several subcategories. We will focus here on a type of learning that is close to the 'trial and error' method called reinforcement learning. The program (which will subsequently be called either agent or system) performs a series of actions and after a certain period of time, the quality of the state reached is evaluated. We then go up in the list of actions carried out and increase their probability of carrying out (if the final evaluation was favorable), or decrease it (if it was unfavorable)
To see the project website:
https://perso.isima.fr/~yjdaniel/cours/projet_etu/projet.html https://yassirmranialaoui.wixsite.com/apprentissage-par-re