Computing PhD Colloquium

Title: Neuromorphic Computing at Scale

Presented by Dr. Tej Pandit, AI Research Scientist, Matrix AI Consortium

Abstract
Large-Scale Neuromorphic Computing systems are designed to mimic the structure and function of the human brain on an unprecedented scale, offering the potential for vastly more energy-efficient and faster processing for AI and complex problem-solving. Neuromorphic platforms often utilize spiking neural networks (SNNs) and non-von Neumann architectures to achieve this efficiency. This paradigm shift could lead to breakthroughs in areas like real-time sensory processing and autonomous systems. These systems represent a fundamental departure from traditional computing, leveraging in-memory computation and event-driven processing for extreme power savings. Their scalability promises to deliver cognitive capabilities far exceeding current deep learning models while operating within tight power budgets.

We introduce the Neuromorphic Commons (THOR) Project, a multi-university U.S. initiative, funded by the National Science Foundation, that provides researchers across various disciplines with open access to a large-scale, heterogeneous neuromorphic computing hardware system to accelerate advancements in bio-inspired AI and computational neuroscience.

Speaker Bio
Dr. Tej Pandit is an AI Research Scientist at the MATRIX AI Consortium. He is focused on designing and establishing an open-access, large-scale heterogeneous neuromorphic hardware ecosystem for the THOR Neuromorphic Commons Project, an initiative that stems from his co-authored Nature publication "Neuromorphic Computing at Scale."

He recently earned his Ph.D. in Electrical Engineering from The University of Texas at San Antonio and holds a Masters in Computer Engineering from the Rochester Institute of Technology.

His research delves into neuromorphic systems to create AI with more natural and continual learning capabilities, exploring how neurogenerative evolutionary mechanisms can be leveraged for advanced AI. His dissertation, "Scalable Continual Learning using Cascading Hypernetworks and Cellular Automata," was honored with the UTSA Outstanding PhD Dissertation Award. He also specializes in designing custom 3D simulation environments to test and refine his AI models.