Detection, Estimation, and Learning
3 credits
Detection, estimation, and learning are underlying to many problems in signal processing, communications, and control engineering and foundational to perform data analysis in various disciplines. This course provides an introduction to guiding principles and classical methods of detection and estimation and to inference on data models. The material covers detection theory, parameter and Bayesian estimation, and machine learning applications in classification, regression and unsupervised learning. The target audience are graduate students from all programs with a solid background in probability, linear algebra, calculus and signals and systems (e.g. obtained or refreshed in the “Mathematical Foundations for Electrical and Computer Engineering” course).
There are three main parts of this course: Detection theory, estimation theory and machine learning methods. The first two follow classical offerings of a detection and estimation course, providing fundamental concepts around the inference of unknown quantities, like the presence of an intruder, the position of an object, the condition of a system, the value of an image pixel, or whether an email is spam. The third part is introductory to machine learning, which can be seen as detection and estimation in the absence of statistical information about the observation model. Applications again include detection, aka classification, and estimation, aka regression. This course structure acknowledges that the commonalities between the statistical signal processing methods of detection and estimation and the learning from data methods in machine learning.