Seminar Announcement
This seminar is sponsored by the IEEE Upper Mon subsection

Abstract: The problem of identification based on biometrics is stated as an M-ary hypothesis testing problem, where M is the number of templates in the data base of an identification system. The data base templates and an input template (to be identified) are modeled as realizations of n consecutive samples of the underlying stationary and ergodic signature random processes with known statistics. Data base templates are assumed to be generated by independent signature random processes. If one of data base templates has a joint distribution with the input template, the input template will be positively identified.

Performance of the identification system is analyzed by applying the theory of large deviations. For a fixed number of hypotheses and increasing number of samples, the minimum probability of error is determined by the smallest component in the vector of large deviation rate functions. For the exponential growth rate R of the number of hypotheses, the identification system is analogous to a communication problem with random coding. The channel reliability function then determines the error rate as a function of the rate of growth R. The capacity of an identification system is one instance of the channel reliability function. The theory is applied to Gaussian and binary-alphabet cases.

Speaker Bio: Natalia A. Schmid received the M.S. degree and the candidate degree in applied physics and mathematics from Moscow Institute of Physics and Technology, Moscow, Russia in 1991 and 1995, respectively. She received the D.Sc. in electrical engineering from Washington University in Saint Louis in 2000. From Fall of 2000 to Spring of 2002 she held postdoctoral positions at Washington University in Saint Louis and at the University of Illinois at Urbana-Champaign.

Presently she is doing independent research and continuing collaboration with her former colleagues from Washington University and the University of Illinois at Urbana-Champaign. Her research interests include regularization methods for statistical recognition, modern estimation and detection theory, statistical signal and image processing, authentication, and information theory. Natalia is a member of IEEE Signal Processing and Information Theory Societies.