Carnegie Mellon University

Applications of High-Dimensional Statistics

Course Number: 47841

Modern data in settings such as e-commerce, healthcare, and operations management are almost always complex and in high dimension. Recent computational and algorithmic advances have led to exciting opportunities to leverage this data for analysis and decision making; indeed, these opportunities have already come to fruition in the form of recommender systems, personalized medicine, and pricing. A critical component of this progress has been a set of probabilistic and statistical tools aimed specifically at high-dimensional settings, including concentration inequalities, minimax theory, and random matrix theory; these tools are precisely the subject of this course.

In many applications, including the aforementioned examples, the data are most naturally represented as a set of matrices or tensors. This course will be largely focused on these settings: we will cover algorithms and major results for matrix estimation, and see recent results in the burgeoning field of tensor estimation. The ultimate goal of the course is to prepare students to apply these same tools in their own research.

Degree: PhD
Concentration: Operations Research
Academic Year: 2019-2020
Semester(s): Mini 3
Required/Elective: Elective
Units: 6


Lecture: 100min/wk and Recitation: 50min/wk