15 095 - Machine Learning Under a Modern Optimization Lens

Description
Prereq: 6.251, 15.093, or permission of instructor. Develops algorithms for central problems in machine learning from a modern optimization perspective. Topics include sparse, convex, robust and median regression; an algorithmic framework for regression; optimal classification and regression trees, and their relationship with neural networks; how to transform predictive algorithms to prescriptive algorithms; optimal prescriptive trees; and robust classification. Also covers design of experiments, missing data imputations, mixture of Gaussian models, exact bootstrap, and sparse matrix estimation, including principal component analysis, factor analysis, inverse co-variance matrix estimation, and matrix completion.
Credits
12
Recent Professors
Schedule Planner
Recent Semesters
Fall 2019
Offered
MW, F
Avg. Sections
2