Event
Booz Allen Hamilton Colloquium: Shantanu Chakrabartty, Washington University of St. Louis
Friday, April 12, 2024
3:30 p.m.
Jeong H. Kim Engineering Building, Room 1110
Darcy Long
301 405 3114
dlong123@umd.edu
Speaker: Shantanu Chakrabartty, Professor, Washington University of St. Louis
Title: Quantum Tunneling, Synaptic Intelligence, and Learning-in-Memory
Abstract: In my talk, I will explore the relationship between the physics of quantum tunneling, synaptic intelligence, and the goal of optimal memory consolidation. I will be discussing a Learning-in-Memory (LIM) paradigm that has been proposed to address performance bottlenecks that arise when training machine learning (ML) and neuromorphic systems. While ML inference involves recall using a fixed or learned set of parameters that can be optimized using compression and sparsification techniques, ML training involves searching over the entire set of parameters, which requires repeated memorization, caching, pruning, and annealing. The emerging compute-in-memory ML architectures can address the memory-wall bottleneck, but they are agnostic to the energy-dissipated due to the number and precision required for the training updates (the update-wall) and when transferring information between short-term and long-term memories (the consolidation-wall). I will demonstrate how a class of quantum tunneling dynamic analog memory can achieve LIM, and how its thermodynamic properties match the physics and energetics of learning.
Bio: Shantanu Chakrabartty is a Clifford Murphey professor and the Vice-dean of Research and Graduate Education in the McKelvey School of Engineering at Washington University in St. Louis. He currently holds an appointment in the department of electrical and systems engineering and the neurosciences division of biology and Biological Sciences at Washington University. Dr. Chakrabartty received his B.Tech degree from Indian Institute of Technology, Delhi and a Ph.D from The Johns Hopkins University. His work covers different aspects of analog computing, and in particular neuromorphic, machine learning, and self-powered systems. Dr. Chakrabartty is a fellow of AIMBE with over 200 journal and conference publications along with twelve issued US patents. Dr. Chakrabartty was a Catalyst Foundation fellow and is a recipient of the NSF CAREER award, the University Teacher-Scholar Award, and the Technology of the Year Award from MSU Technologies.
