Scalability in Machine Learning

12 units · Graduate

Fall: 12 units The goal of this course is to provide a survey into some of the recent advances in the theory and practice of dealing with scalability issues in machine learning. We will investigate scalability issues along the following dimensions: Challenges with i) large datasets, ii) high-dimensions, and iii) complex data structure. The course is intended to prepare students to write research papers about scalability issues in machine learning. This is an advanced-level, fast-paced course that requires students to already have a solid understanding of machine learning (e.g. by taking an intro to ML class), good programming skills in Python, and being comfortable with dealing with abstract mathematical concepts and reading research papers. The course will have significant overlap with 10-405 /605/805, but 10-745 will be faster-paced and go deeper into the theoretical investigations of the methods. Some of the classes will be flipped that will require students to watch a video lecture or read a research paper before the class, and the content will be discussed during the class time. The class will include a course project, HW assignments, and two-in class exams. Prerequisites: 10-715 Min. grade B or 10-401 Min. grade B or 10-601 Min. grade B or 10-701 Min. grade B or 10-301 Min. grade B or 10-315 Min. grade B