This course focuses on Deep Reinforcement Learning (DRL), a relatively new area of computer science that has driven many recent advancements in artificial intelligence. DRL has been successfully applied to create the first computer programs capable of defeating humans in a wide range of classic board games (e.g., Chess, Go, Shogi), mastering numerous video games, and developing advanced robotic control systems, including self-driving cars. DRL is also being used to tackle biological challenges such as protein folding, create new materials, and contribute to mathematical discovery. Additionally, DRL plays a key role in powering the intelligence behind modern large language models like ChatGPT. Students will be expected to have completed a course covering the basics of neural networks using some existing framework (e.g., Keras, PyTorch, JAX, TensorFlow), although we will begin with a review of this.
