Artificial intelligence and reinforcement learning

Dátum
Szerzők
Khedr, Mohamed
Folyóirat címe
Folyóirat ISSN
Kötet címe (évfolyam száma)
Kiadó
Absztrakt
In my thesis I would like to discuss AI (Artificial Intelligence), its connection to RL (Reinforcement Learning), and how to incorporate these ideas into a game. With the aid of helpful Python libraries like pygame and PyTorch, I have utilized Python as a programming language in Visual Studio Code to help us construct the Snake Game. I developed an agent (AI) utilizing the aforementioned notions that can learn, play the snake game, and get a score of 75 all by itself. Two ideas were proposed that are a mixture of this method, namely the CNN (Convolution Neural Network) and Q-learning algorithm, by employing the Deep Q-learning algorithm, a model-free RL algorithm. I described the method we utilized and how the snake practices before playing the game in order to find the food. Lastly, comparing the outcomes of the agent utilizing various activation functions and doing away with a particular reward function.
Leírás
Kulcsszavak
Reinforcement Learning, Python, Game
Forrás
Gyűjtemények