When: Wednesday, November 1, 2017
Where: PGH 563
Time: 11:00 AM – 12:30 PM

Correlation Study on Attributes of Unsteady Flows

Speaker: Marzieh Berenjkoub

In this work, we extend the existing spatial correlation quantification metrics, i.e. the Local Correlation Coefficient (LCC), to spatio- temporal domain, and introduce a new metric for the quantification of the correlation of vector-valued attributes for unsteady flows. In addition, we compute the mutual information between pairs of attributes over time to analyze their temporal correlations. We have applied our correlation quantification metrics to a number of 2D and 3D unsteady flows to assess its usefulness.

Bio:

Marzieh Berenjkoub is a Ph.D. student at the University of Houston, studying computer science and scientific visualization under Professor Guoning Chen. Her research interests include data visualization, large-scale scientific simulations, and high performance computing and computer graphics. Berenjkoub received a BS in Software Engineering from University of Isfahan, Iran and MS in Computer Engineering from University of Tabriz, Iran.

SonicDoor: Scaling Person Identification with Ultrasonic Sensors by Novel Modeling of Shape, Behavior and Walking Patterns

Speaker: Nacer Khalil

We propose a new method to identify occupants by sensing their body shape, movement, and a set of walking behaviors as they walk through a door that is instrumented with three ultrasonic sensors. The proposed method is able to identify up to 100 occupants with an accuracy of 90.2%, which makes it suitable for large-scale realistic buildings.

Bio:

Nacer Khalil is a fifth year PhD Candidate advised by Professor Omprakash Gnawali. He is interested in sensor networks, signal processing, and machine learning.