Spintronic Bayesian Hardware Driven by Stochastic Magnetic Domain Wall Dynamics
This addresses the need for reliable, uncertainty-aware AI in safety-critical domains by offering a scalable hardware accelerator, though it is incremental as it builds on existing probabilistic computing concepts with a novel physical implementation.
The paper tackles the challenge of implementing probabilistic neural networks (PNNs) for uncertainty estimation in AI by introducing a Magnetic Probabilistic Computing (MPC) platform that leverages stochastic magnetic domain wall dynamics, achieving a seven orders of magnitude improvement in figure of merit over CMOS implementations on CIFAR-10 classification tasks.
As artificial intelligence (AI) advances into diverse applications, ensuring reliability of AI models is increasingly critical. Conventional neural networks offer strong predictive capabilities but produce deterministic outputs without inherent uncertainty estimation, limiting their reliability in safety-critical domains. Probabilistic neural networks (PNNs), which introduce randomness, have emerged as a powerful approach for enabling intrinsic uncertainty quantification. However, traditional CMOS architectures are inherently designed for deterministic operation and actively suppress intrinsic randomness. This poses a fundamental challenge for implementing PNNs, as probabilistic processing introduces significant computational overhead. To address this challenge, we introduce a Magnetic Probabilistic Computing (MPC) platform-an energy-efficient, scalable hardware accelerator that leverages intrinsic magnetic stochasticity for uncertainty-aware computing. This physics-driven strategy utilizes spintronic systems based on magnetic domain walls (DWs) and their dynamics to establish a new paradigm of physical probabilistic computing for AI. The MPC platform integrates three key mechanisms: thermally induced DW stochasticity, voltage controlled magnetic anisotropy (VCMA), and tunneling magnetoresistance (TMR), enabling fully electrical and tunable probabilistic functionality at the device level. As a representative demonstration, we implement a Bayesian Neural Network (BNN) inference structure and validate its functionality on CIFAR-10 classification tasks. Compared to standard 28nm CMOS implementations, our approach achieves a seven orders of magnitude improvement in the overall figure of merit, with substantial gains in area efficiency, energy consumption, and speed. These results underscore the MPC platform's potential to enable reliable and trustworthy physical AI systems.