Speaker

Rabab Abdelfattah, University of South Carolina

Abstract

Artificial Intelligence (AI) is a rapidly growing field that has sprawled into dozens of applications, such as computer vision, traffic prediction, flight safety for unmanned aerial vehicles (UAVs), self-driving cars, and power grid inspections. For instance, AI plays a key role in precise scene understanding for real-time autonomous UAV-based power grid applications, enabling UAV tracking, aerial monitoring, inspection, and flight safety. Additionally, AI is used to address background and object complexity in scene recognition. AI evolution has been concurrent with technology scaling, leading to the integration of more than two billion transistors onto a single chip. As a result, Networks-on-Chip (NoCs) have emerged as the most competent communication fabric to connect an increasing number of processing/memory elements. NoC characteristics are significantly influenced by the network topology, data transfer control method, and routing algorithm. This presentation will discuss two selected projects in AI and NoC topics. The first project aims to detect power lines from aerial images captured by UAVs. The second project proposes a routing algorithm based on process variation for NoC routers. Combining AI and NoC knowledge paves the way for AI accelerators to accelerate AI algorithms while saving power consumption.

Biography

Rabab Abdelfattah is a Ph.D. candidate at the University of South Carolina. Prior to her current Ph.D. program, Rabab earned her master's and first Ph.D. degrees in computer engineering. Her current research interests include NoC/system-on-chip, low-power techniques, process variation, NoC algorithms, computer vision, deep learning, and AI. Rabab's work has been published in international conferences and journals in computer science and microelectronic systems, including ACCV, BMVC, ISCAS, and IEEE Transactions on Very Large-Scale Integration Systems Journal. Additionally, she has authored two books and a book chapter. Rabab is a reviewer in top computer engineering/science conferences and journals such as CVPR, ACCV, ECCV, IEEE Transactions on Very Large-Scale Integration Systems, Pattern Recognition Letters Journal, Journal of Circuits, Systems, and Computers, Microprocessors and Microsystems Journal, and Knowledge-Based Systems Journal.

Details

Wednesday, May 10, 2023, 1:00 – 2:15 pm

Hybrid event:

Available In-person: College of Engineering, Room 1507 (Engineering Auditorium)
Available Online: ZOOM
Meeting ID: 938 4652 7736  Passcode: 502688