Matrix Methods for Signal Processing, Data Analysis and Machine Learning

4 credits · College of Engineering catalogue

Advisory Prerequisite: EECS 351 or graduate standing. Enforced Prerequisite: None. Credit Exclusions: Student may only receive credit for one course , ECE 505 and ECE 551. (4 credits) Theoretical topics include subspaces, eigendecomposition, singular value decomposition, projection theorem, norms, constrained, regularized, unconstrained least-squares fitting, Procrustes method, low-rank approximation, matrix completion, and basic optimization methods. Applications include handwritten digit recognition, image deblurring, ranking of webpages, image compression, social networks, recommender systems and robust regression.