Probabilistic Computing with p-bits: Optimization, Machine Learning and Quantum Simulation

9th October 2025

Timing : 1 pm ET

Please use this zoom link for joining the webinar

For a list of all talks at the NanoBio seminar Series Fall'25, see here

Positioned at the intersection of classical and quantum computing, probabilistic computing provides a physics-inspired approach to domain-specific computation by harnessing the inherent randomness of devices such as stochastic magnetic tunnel junctions (sMTJs) [1-5]. These devices naturally generate tunable randomness and replace thousands of transistors per p-bit, reducing energy consumption. In this talk, we demonstrate how networks of probabilistic bits (p-bits) in both digital CMOS and hybrid CMOS + sMTJ implementations can accelerate tasks in optimization, machine learning, and statistical inference through asynchronous updates and sparse connectivity. We also describe a distributed FPGA-based prototype that achieves near-linear speedup with minimal overhead, enabling the study of large-scale problems. In particular, we will highlight recent results on 3D Spin Glass systems benchmarked against quantum annealers, showing how targeted architectures can drastically reduce time and energy to solution. The unique combination of intrinsic randomness in sMTJs and flexible CMOS circuits may pave the way for next-generation probabilistic computers that can address computationally intensive, previously intractable tasks in a wide range of applications.

Dr. Kerem Camsari
Associate Professor

Kerem Y. Camsari is an Associate Professor of Electrical and Computer Engineering at UC Santa Barbara. He received his Ph.D. from Purdue University in 2015 and continued there as a postdoctoral researcher until 2020. His research spans quantum transport and probabilistic computing, where he helped introduce p-bits and p-circuits as a practical bridge between classical and quantum hardware. His work has been published in Nature, Nature Electronics, Science Advances, and Physical Review X, and he has delivered over 50 invited talks at venues including APS, DRC, MMM, IEDM, and VLSID. He has served on program committees for DATE, ICRC, and IEDM, and currently leads the unconventional computing section of the IEEE Nanotechnology Council’s technical committee. Kerem’s honors include the Bell Labs Prize, Misha Mahowald Prize, IEEE Magnetics Society Early Career Award, ONR YIP, and NSF CAREER. He was an IEEE Distinguished Lecturer in 2024 and is a senior member of IEEE.