In this class, we’ll develop ways to think about and solve probabilistic inference problems with robots. Each week we'll introduce a new real-world challenge, and students will work in teams to reason about the math behind the task, design a method, and implement it in on physical robots! Through this process, we’ll actually arrive at methods that are both fundamental and very modern, used as the basis for contemporary artificial intelligence, machine learning, signal processing and communication. The course material will build from basic principles of probability to concepts such as hypothesis testing, the EM algorithm, and Bayesian filtering. In parallel we'll develop methods for robot localization, optimal control, and planning. The class will have minimal lecture, and will mostly consist of round-table discussion and programming with real robots.

An introduction to nuclear weapons design, including relatively recent developments. Discusses the involved physics, chemistry, and engineering. Additionally, the course will address both the science and impact of nuclear reactors, enrichment, fusion, and ethics.

An instructive, practical course on how to build a nuclear bomb from the ground up. Includes a discussion of the involved physics, chemistry, and engineering. By the completion of the course, students should have a fundamental understanding sufficient to construct a small warhead just using leftovers from the previous evening's dinner. Additional, students will be able to apply these principles in understanding modern nuclear reactors, and will come out with a historical and ethical understanding of the nuclear age.

Learn how computers learn. Starts at the mathematical and intuitive basis for inference, and builds up to a practical exploration of deep learning with real datasets. A project + discussion hybrid class, fun for the whole family.

A practical lab-based course on how computers can learn to solve complex problems. In addition to lecture sessions, students will develop their own algorithms that tackle real world problems with real datasets. An emphasis will be placed on developing a strong intuition for the optimization methods that underlie the machine learning applications transforming the world today. The course will cover topics such as both constrained and unconstrained optimization methods, numeric and combinatorial optimization, and the classifiers and regressors that can be built upon these frameworks.

Programming is an important skill to learn for modern life. It is applicable in basically every field now. This course is going to teach coding from the very basics assuming that students have little to no knowledge about programming. Students will learn to code in python by writing bots to play a simplified version of Battlecode, MIT's oldest and largest programming competition. We will start simple with the basic ideas of imperative programming languages, before explaining how loops, functions, and classes work. At the end of the class there will be a tournament to see who wrote the best bot.

When we think about artificial intelligence, we often study the key: techniques like machine learning, neural networks, and bayesian nonparametrics which allow us to tackle more and more complex challenges. However, we often neglect to study the lock. In this short class, we will explore classifications of different types of computational tasks, and learn what exactly computers can do.