Dr. Andrew Drozd, President & Chief Scientist of ANDRO Computational Solutions, LLC, has been invited as a panelist at The 36th IEEE International Workshop on Communications Quality and Reliability (IEEE-CQR).
Dr. Drozd will join David Lu (AT&T) and Mallik Tatipumala (CTO, Ericsson Silicon Valley) on September 14th at the in person conference in Arlington, VA. Time of the panel TBD. View full panel details here
Panel: Machine Learning for Real-time Network Optimization – What Can Go Wrong?
Panel Moderator: Carol Davids, Research Associate Professor, CS Department and Founding Director, Real Time Communications Laboratory, Illinois Institute of Technology
Panel Description: Data and telecommunications networks produce enormous volumes of operational data. Researchers today are exploring ways to apply ML/AI techniques to these data to develop self- regulating networks that can predict and respond to network events and trends in real-time. As the network traffic increases in volume and complexity, and new networked applications place new demands, we can expect that the research of today will produce solutions that the network operators will be eager to adopt. These applications could prove very helpful. What do we need to do to avoid the problems that applications created using ML/AI techniques have produced in other areas?
Self-driving cars, for example, demonstrate that when their ML/AI-developed applications fail, they fail in spectacular and deadly ways, magnifying their errors rapidly and without self-correction. Applications to determine the length of jail terms, the amount of bail, the terms and conditions of loans, university admissions, and hiring choices have tended to magnify social biases rather than help eliminate them. Applications created using ML/AI techniques and applied to the output of surveillance cameras have been unable to identify women with dark skin as even humans!
Can applications that aim to provide real-time network management fail in similar ways? We look at different types of potential failures: Technical failures in which a small wrong decision might escalate bringing a network down or amplifying rather than correcting the initial problem, and also the Societal failures: Could low numbers of Internet users in an underserved community, result in fewer network resources allocated to that community, amplifying rather than improving the available service?
These and related concerns will be addressed by our panelists, a diverse group of experts from the research and development, service provider, vendor, legal and advocacy sectors.
The 36th IEEE International Workshop on Communications Quality and Reliability (IEEE-CQR), will be held in-person in Arlington, VA, September 13-15, 2022. All speakers and most registrants will be on-site. Please see the registration page (https://cqr2022.ieee-cqr.org/registration/) for the various options. The workshop offers technical sessions, original paper presentations, and keynotes panels designed to further career opportunities and the in-depth understanding of key issues impacting communications networks quality and reliability.