Machine Learning: Theory App

3 Credits · Level: Graduate, Undergraduate

Machine Learning describes a collection of computational techniques for finding patterns in data. This course focuses on supervised machine learning, which deals with using historical labeled data to construct a predictor to accurately label future data. We will cover how to formulate the supervised learning problem in mathematical terms, describe a measure of performance, restrict a search space for constructing models for prediction, optimize performance over the search space, and check for generalization. We will also discuss statistical learning guarantees and of the limits in such results. The course, while theory-centric, will provide substantive exposure to implementation through programming exercises, and will establish foundations for constructing architectural ingredients of neural networks, which are ubiquitous in modern machine learning practice.