Talk: Dr. Mohamed Ibrahim "Biologically Inspired Neuro-Symbolic Computing for Human-Centered AI"
Monday, February 3, 2025 · 11 AM - 12 PM
Short Speaker Bio: Mohamed Ibrahim is a Senior Research Engineer at Georgia Tech with 14 years of combined experience across industry and academia. He has previously worked as a postdoctoral researcher at UC Berkeley and as an SoC Design Engineer at Intel Corporation. Dr. Ibrahim earned his Ph.D. in Electrical and Computer Engineering (ECE) from Duke University, where he spearheaded multiple cross-departmental research initiatives. His research focuses on advancing human-centered AI through biologically inspired computing and algorithm-system-hardware co-design and optimization. His work has been recognized with a Best Paper Award at DATE’17, as well as the 2018 Council of Graduate Schools/ProQuest Distinguished Dissertation Award and Duke ECE’s 2019 Outstanding Dissertation Award. Learn more about him at https://mohamed-s-ibrahim.github.io.
Abstract: Human-centered AI has the potential to revolutionize fields like personalized healthcare, extended reality applications, and autonomous systems by enabling machines to better understand and collaborate with people. In this talk, I will explore how neuro-symbolic computing can advance AI systems to be more aligned with human needs and capabilities. I will introduce core concepts in neuro-symbolic AI, with an emphasis on vector-symbolic computing, and discuss how incorporating perception, reasoning, and action can enhance the responsiveness, interpretability, and adaptability of AI systems. At the heart of my research is an interdisciplinary methodology that combines human-centered AI with neuro-symbolic computing and algorithm-system-hardware co-optimization. I will illustrate this approach through a key project: the Hyperdimensional Processing Unit (HPU), a general-purpose processor optimized for vector-symbolic computing, with brief overviews of its dataflow and control flow. Finally, I will outline my vision for future research to further enhance human-centered AI through biologically inspired designs and efficient computing architectures.
