Self-similar Magneto-electric Nanocircuit Technology for Probabilistic Inference Engines

Abstract

Probabilistic graphical models are powerful mathematical formalisms for machine learning and reasoning under uncertainty that are widely used for cognitive computing. However they cannot be employed efficiently for large problems (with variables in the order of 100K or larger) on conventional systems, due to inefficiencies resulting from layers of abstraction and separation of logic and memory in CMOS implementations. In this paper, we present a magneto-electric probabilistic technology framework for implementing probabilistic reasoning functions. The technology leverages Straintronic Magneto-Tunneling Junction (S-MTJ) devices in a novel mixed-signal circuit framework for direct computations on probabilities while enabling in-memory computations with persistence. Initial evaluations of the Bayesian likelihood estimation operation occurring during Bayesian Network inference indicate up to 127x lower area, 214x lower active power, and 70x lower latency compared to an equivalent 45nm CMOS Boolean implementation.

References in corpus (1)

• Energy-efficient magnetoelastic non-volatile memory

Cited by in corpus (13)

• Low Barrier Nanomagnets as p-bits for Spin Logic
• Straintronics: Manipulating the Magnetization of Magnetostrictive Nanomagnets with Strain for Energy-Efficient Applications
• A building block for hardware belief networks
• Error Probability in Magneto-elastic Switching of Non-ideal Nanomagnets with Defects: A Case Study for the Viability of Straintronic Logic and Memory
• Energy-efficient switching of nanomagnets for computing: Straintronics and other methodologies
• Low Barrier Nanomagnet Design for Binary Stochastic Neurons: Design Challenges for Real Nanomagnets with Fabrication Defects
• Dynamic Error in Strain Induced Magnetization Reversal of Nanomagnets due to Incoherent Switching and Formation of Metastable States: A Size-dependent Study
• Applications of Nanomagnets as Dynamical Systems
• Nanomagnetic Boolean Logic -- The Tempered (and Realistic) Vision
• A Non-Volatile All-Spin Non-Binary Matrix Multiplier: An Efficient Hardware Accelerator for Machine Learning
• Quick and energy-efficient Bayesian computing of binocular disparity using stochastic digital signals
• Polarized Neutron Reflectometry Study of Depth Dependent Magnetization Variation in Co Thin Film on a PMN-PT Substrate
• Straintronic magneto-tunneling-junction based ternary content addressable memory