Optimization@MIT

2.687 -- Time Series Analysis and System Identification
Course Description: Matched filtering, power spectral estimation and adaptive signal processing and system identification algorithms are introduced. Algorithm development is framed as an optimization problem, and methods of finding both optimal and approximate solutions are described. Introduction to time-varying systems, first and second moment characterizations of stochastic processes, and state-space models. Algorithm derivation, performance analysis and robustness to modeling errors are covered for matched filter and power spectral estimation algorithms, stochastic gradient algorithms (LMS and its variants), Least Squares algorithms (RLS, order- recursive approaches), and the discrete-time Kalman Filter and its derivatives. Includes laboratory exercises involving working with experimental data from a variety of fields. Term paper/project is required.

This class is at the Graduate level
Instructor: J. C. Preisig
Prerequisites: 6.003, 6.431, 18.06

Back to Classes

Suggestions, feedback? Please send your thoughts to optimization-www@mit.edu.