IEEE Communication Society New York Chapter, Technical Seminar Series
The IEEE Communications Society (ComSoc) promotes the advancement of science, technology and applications in communications and related disciplines. It fosters presentation and exchange of information among its members and the technical community throughout the world. The Society maintains a high standard of professionalism and technical competency. The IEEE Communications Society is a professional society of the IEEE.
The IEEE ComSoc New York Chapter is launching a series of technical seminars for the New York area IEEE members and the general public. We will invite researchers and professionals to share their latest work on a variety of topics in communications and related areas. This is the first seminar of the series.
Date and Time
Location
Hosts
Registration
- Date: 30 Nov 2020
- Time: 06:00 PM to 08:00 PM
- All times are (UTC-05:00) Eastern Time (US & Canada)
-
Add Event to Calendar
- Starts 17 November 2020 05:18 PM
- Ends 30 November 2020 05:18 PM
- All times are (UTC-05:00) Eastern Time (US & Canada)
- No Admission Charge
Speakers
Yun Ye of City University of New York
Scalable Disparity Estimation in 3D Multimedia Communications
This talk introduces a scalable disparity estimation method using stereoscopic image pairs from binocular cameras. Traditional grid-based multiresolution methods consider local features of the initial supporting points on the sparse grid, and fall short of resolving spurious matches caused by similar textures, which are commonly encountered in an indoor environment with reduplicate building structures or wall decorations. This will in turn result in incorrect estimation on the finer grid. In addition, the disparity values on the finer grid estimated using the local features could not be further utilized to estimate the disparity of the neighboring pixels. Instead we performed global optimization on the initial supporting points. A best disparity for each initial supporting point is selected from multiple candidates through Iterative Conditional Mode. The disparity estimation process on the finer grid is formulated as an energy minimization problem. Both the data consistency term and the smoothness term are included in the iterative solution procedure implemented in a course-to-fine manner. Its application in delay-sensitive 3D multimedia communications will be demonstrated with the example of video object tracking.
Biography:
Yun Ye (M’ 15) received the Ph.D. degree in Computer Engineering from University of Nebraska-Lincoln, NE, USA, in 2013. She is currently an Associate Professor with Department of Mathematics, Engineering and Computer Science, LaGuardia Community College at City University of New York, NY, USA. She was Chair of IEEE ComSoc NY Chapter from 2017 to 2019. Her research interests include cross-layer design in secure and energy efficient wireless communications, video surveillance, and 3D multimedia signal processing.
Jiang Zhu of Zhejiang University
Approximate Bayesian Algorithms Design for Generalized Linear Models: A Modularized Point of View
A signal undergoes a linear transform followed by a componentwise nonlinear transform is referred to as the generalized linear models (GLMs), and substantial examples such as image classification, parameter estimation from quantized data and phase retrieval can be formulated as a GLM problem. A well-known algorithm for solving the GLM is the generalized approximate message passing (GAMP). In this talk, designing approximate Bayesian algorithms from modularized point of view especially for the GLM is presented. Note that for the standard linear model (SLM), many existing approximate Bayesian inference algorithms such as (approximate message passing (AMP), vector approximate message passing (VAMP), sparse Bayesian learning (SBL), variational Bayesian inference (VBI) have been developed. Compared to the standard linear models (SLMs), solving the GLMs is more challenging because of the coupling of the linear and nonlinear transforms. Here an expectation propagation (EP) approach is utilized to show the relationship between the recovery algorithms of SLMs and that of GLMs, which provides insight into the designing of Bayesian algorithms for GLMs based on the existing Bayesian algorithms. According to EP, the GLM can be iteratively approximated as a sequence of SLM subproblems, and thus the standard Bayesian algorithms can be easily extended to solve the GLMs. In addition, substantial examples are also provided to demonstrate the effectiveness of the approach. This talk is based on joint work with Xiangming Meng and Sheng Wu.
Biography:
Jiang Zhu received B.E. from Harbin Engineering University in 2011, Harbin, China, and Ph.D. degree in electronic engineering from Tsinghua University, Beijing, China, in 2016. From Feb. 2015 to Aug. 2015, he was a visiting student with the signal processing and communication laboratory, Lehigh University. Since 2016, he joined Ocean College, Zhejiang University as a lecturer. Dr. Zhu is a member of IEEE. His current research interests include statistical signal processing, compressed sensing and unlabeled sensing.