APPLIED MATH SEMINAR

Speaker: Neta Rabin, Yale University

Title: Detection and tracking the behavior of dynamical systems

When/where: Tuesday, October 20th, 4:15 PM, AKW 200

Abstract:
Finding meaningful geometric descriptions in large heterogeneous
datasets that come from real-life applications is an important task. In many of
these application the collected data is dynamic and the tracked process
constantly changes with time. The goal is to identify trends in the process
that deviate from normal behavior.  The high-dimensional dataset, which
describes the measured/observed parameters of a dynamical system, is embedded
into a lower-dimension space by the application of the  Diffusion Maps. Then, a
clustering process or detection of the the sought after anomalies is performed
in this
lower-dimension embedded space.
In this talk I will present the results of these developed methodologies  to
several different dynamical systems  such as transaction based system,
acoustics detection, image classification and medical data.