Reinforcement Learning: Pac-Man

Ever wondered how the best chess players are being beat by machines? Or how cars are able to learn how to drive by themselves?

Sit back as we demonstrate reinforcement learning by building an agent to beat Pac-Man. In addition to using the traditional method called Q-learning, we will make use of deep Q-learning, which utilizes neural networks. Along the way, will describe the techniques and libraries used, and also demonstrate the agent through the learning process. During your time with us, you will gain insight into reinforcement learning theory, how you can build deep learning models with libraries, such as Keras, and see how this culiminates in a Pac-Man-beating agent.

Prior knowledge of machine learning is not necessary to enjoy the presentation, but you are most likely to understand more if you have dealt with the basics of machine learning previously.