International Program on Machine Learning
Program Overview
Even though there are plenty of machine learning courses and programs around, there are few systematic programs on the theoretical foundation of machine learning and its application to science, which are two of the most promising new directions in AI. In addition, the educational resources in AI are very unevenly distributed over the world, with many countries lacking access to the cutting edge of these vastly important scientific areas.
As the first certificate program in PKU's Global Open Courses Program (此处增加全球课堂的网页链接), the International Program on Machine Learning (IPML) is designed to provide international students with a rare opportunity to learn and conduct research in theoretical machine learning and AI for Science. The program is organized by the Center for Machine Learning Research (CMLR) at Peking University (web link), a leading institution in these areas. Its goal is to provide students with the necessary background knowledge leading towards pursuing a career in exploring the frontier of theoretical machine learning and AI for Science, as well as the opportunity to work with the world’s leading experts in these areas. IPML offers both online courses and in-person research opportunities.
Program Description
The program comprises the following:
1. Two online courses during Fall 2022 and Spring 2023. Those students who successfully complete the two online courses will receive a program certificate from the International Program in Machine Learning.
2. A summer residential learning and research experience at PKU during July/August 2023. A certain number of full scholarships to attend this summer camp will be provided for high achieving students who have successfully taken the two online courses.
3. The top students from the summer program will qualify to receive a full scholarship to begin a graduate program at PKU in the new field of machine learning.
Course Description
1. Mathematical Introduction to Machine Learning (Fall 2022, Date: September 5, 2022 – January 8, 2023)
This course gives a mathematical introduction to the most important areas in machine learning, particularly deep learning – based machine learning. The basic starting point is to think about supervised learning, unsupervised learning and reinforcement learning as problems about approximating functions, approximating probability distributions and solving Bellman equations, respectively. Relevant issues in approximation theory, estimation error and the properties of training algorithms will be discussed. The course will also discuss how to use mathematical tools, such as differential equations, to formulate machine learning models.
2. Topics in AI for Science (Spring 2023, Date: February 20 – June 25, 2023)
Taught by Bin Dong and Weinan E
This course introduces students to the frontier of a fascinating new field: AI for Science, namely, how machine learning and other AI techniques can be used to advance the frontier of science. Topics to be covered include:
(1) Introduction to the relevant physical models used in science.
(2) Introduction to machine learning.
(3) Machine learning – based algorithms for the quantum many-body problem.
(4) Machine learning – based molecular modeling.
(5) Machine learning – based algorithms for PDEs.
(6) Application areas include protein folding, drug design, combustion, solid mechanics, fluid mechanics, turbulence modeling, computational imaging, etc.
Students will be assigned projects on specific topics in AI for Science to help bring them to the frontier of this field. They may work in groups based on their research interests. At the end of the course, each participating student must submit a term paper, and his/her performance will be evaluated based on the quality of the term paper. The term paper needs to demonstrate a good understanding of some aspects of the field and present some suggestions for future research directions. One specific aim of this course is to guide participating students to find a particular topic of their interest in AI for Science. Students may conduct further in-depth research on the topic during the summer research program.
Summer Camp and Graduate Program
A summer residential learning and research experience at PKU will follow the two aforementioned IPML courses, in July/August 2023. A certain number of full scholarships to attend this summer camp will be provided for high achieving students who have successfully taken the two online courses. Top students from the summer program will qualify to receive a full scholarship to begin a graduate program at PKU in the new field of Machine Learning.
(Note: The online courses, summer residential camp and graduate program will be conducted in English.)