Seminar in applied mathematics and statistics
SPEAKER: Amit Mitra (IIT Kanpur, India)
TITLE: Order Estimation in Nonlinear Signal Processing Models
ABSTRACT: A fundamental problem of interest in digital signal processing and time series analysis is order estimation of nonlinear models. Signal processing has a wide variety of applications, ranging from radar and sonar systems to computer assisted medical diagnostics, wireless communications, tracking of objects in space, texture classification, speech synthesis, meteorology and geophysics. Parametric methods of signal processing requires the estimation of integer-valued parameters that typically specify the noise contaminated nonstationary signal model. This detection problem is an important and challenging problem in signal processing as the vector of real and/or complex valued parameters characterizing the signal can only be estimated after the order is estimated efficiently. We present in this talk some recent novel likelihood based methods for order estimation of important signal processing models. We discuss large sample asymptotic properties of the resultant estimators. Finite sample simulation studies and real data analysis will also be presented comparing the performance of the order estimation methods.
Friday, May 24 at 14.15: Sharmishtha Mitra
Wednesday, May 29 at 15.15: Frank van der Meulen
Tuesday, July 2 at 13.15: Gaurav Khemka