A Novel Compound Synapse Using Probabilistic Spin–Orbit-Torque Switching for MTJ-Based Deep Neural Networks

@article{Ostwal2019ANC,
  title={A Novel Compound Synapse Using Probabilistic Spin–Orbit-Torque Switching for MTJ-Based Deep Neural Networks},
  author={Vaibhav Ostwal and Ramtin Zand and Ronald F. Demara and Joerg Appenzeller},
  journal={IEEE Journal on Exploratory Solid-State Computational Devices and Circuits},
  year={2019},
  volume={5},
  pages={182-187}
}
Analog electronic nonvolatile memories mimicking synaptic operations are being explored for the implementation of neuromorphic computing systems. Compound synapses consisting of ensembles of stochastic binary elements are alternatives to analog memory synapses to achieve multilevel memory operation. Among the existing binary memory technologies, magnetic tunneling junction (MTJ)-based magnetic random access memory (MRAM) technology has matured to the point of commercialization. More importantly… 

Figures from this paper

Gradient Descent on Multilevel Spin–Orbit Synapses with Tunable Variations
TLDR
The gradient descent algorithm, a primary learning algorithm, implemented on a 2 × 1 SOT synaptic array is reported and it is found that the classification accuracies can be improved dramatically by decreasing the cycle‐to‐cycle variation.
Implementation of an efficient magnetic tunnel junction-based stochastic neural network with application to iris data classification.
TLDR
Investigation of how crucial parameters, such as stochastic bitstream length, number of hidden layers and number of nodes in a hidden layer, need to be set precisely to realize an efficient MTJ-based Stochastic neural network (SNN) can prove a promising alternative for highly efficient digital stoChastic computing applications.
Magnetic Elements for Neuromorphic Computing
TLDR
A review of diverse approaches to implement neuromorphic functionalities in novel hardware using magnetic elements, published during the last years underlines the possible applications of magnetic materials and nanostructures in neuromorphic systems.
Neuromorphic Engineering: From Biological to Spike‐Based Hardware Nervous Systems
TLDR
Fundamental knowledge related to the structures and working principles of neurons and synapses of the biological nervous system is reviewed and an overview is provided on the development of neuromorphic hardware systems, from artificial synapses and neurons to spike‐based neuromorphic computing platforms.
A survey of in-spin transfer torque MRAM computing
TLDR
This study reviews state-of-the-art techniques for managing IMC with an emphasis on spin-transfer torque-MRAM computing via design schemes at the bit-cell, circuit, and system levels and demonstrates the existing limitations of in- MRAM computing and potential methods for overcoming these issues.
Spintronic memristors for neuromorphic circuits based on the angular variation of tunnel magnetoresistance.
In this study, a new type of compact magnetic memristor is demonstrated. It is based on the variation of the conductivity of a nano-sized magnetic tunnel junction as a function of the angle between
Interconnect Parasitics and Partitioning in Fully-Analog In-Memory Computing Architectures
TLDR
This paper investigates the effect of wire parasitic resistance and capacitance on the accuracy of DNN models deployed on fullyanalog IMC architectures, and proposes a partitioning mechanism to alleviate the impact of the parasitic while keeping the computation in the analog domain through dividing large arrays into multiple partitions.
Roadmap of Spin–Orbit Torques
Spin–orbit torque (SOT) is an emerging technology that enables the efficient manipulation of spintronic devices. The initial processes of interest in SOTs involved electric fields, spin–orbit
Adaptive Non-Uniform Compressive Sensing Using SOT-MRAM Multi-Bit Precision Crossbar Arrays
TLDR
The Adaptive Compressed-sampling via Multi-bit Crossbar Array (ACMCA) approach to intelligently generate the CS measurement matrix using a multi-bit SOT-MRAM crossbar array is proposed.
Adaptive Non-Uniform Compressive Sensing using SOT-MRAM Multibit Crossbar Arrays
TLDR
The Adaptive Compressed-sampling via Multibit Crossbar Array (ACMCA) approach to intelligently generate the CS measurement matrix using a multibit SOT-MRAM crossbar array is proposed.
...
1
2
...

References

SHOWING 1-10 OF 30 REFERENCES
All Spin Artificial Neural Networks Based on Compound Spintronic Synapse and Neuron
TLDR
It has been demonstrated that the proposed compound spintronic device can achieve designable and stable multiple resistance states by interfacial and materials engineering of its components.
Stochastic spintronic device based synapses and spiking neurons for neuromorphic computation
TLDR
A compound scheme is firstly proposed, which employs multiple MTJs connected in parallel operating in the stochastic regime to jointly behave a single synapse, aiming to achieve an analog-like weight spectrum.
Spin-Transfer Torque Magnetic Memory as a Stochastic Memristive Synapse for Neuromorphic Systems
TLDR
It is shown that when used in a non-conventional regime, STT-MTJs can additionally act as a stochastic memristive device, appropriate to implement a “synaptic” function in robust, low power, cognitive-type systems.
Ferroelectric FET analog synapse for acceleration of deep neural network training
TLDR
A transient Presiach model is developed that accurately predicts minor loop trajectories and remnant polarization charge for arbitrary pulse width, voltage, and history of FeFET synapses and reveals a 103 to 106 acceleration in online learning latency over multi-state RRAM based analog synapses.
A compound memristive synapse model for statistical learning through STDP in spiking neural networks
TLDR
The compound memristive synapse may provide a synaptic design principle for future neuromorphic architectures because its emergent synapse configuration represents the most salient features of the input distribution in a Mixture-of-Gaussians generative model.
Chain of magnetic tunnel junctions as a spintronic memristor
In the context of neuromorphic computation, spintronic memristors are investigated for their use as synaptic weights. In this paper, we propose and experimentally demonstrate a resistive synaptic
Composable Probabilistic Inference Networks Using MRAM-based Stochastic Neurons
TLDR
The MNIST dataset is leveraged to investigate the energy and accuracy tradeoffs of seven distinct network topologies in SPICE using the 14nm HP-FinFET technology library with the nominal voltage of 0.8V, in which an MRAM-based neuron is used as the activation function.
Neuromorphic computing with nanoscale spintronic oscillators
TLDR
It is shown experimentally that a nanoscale spintronic oscillator can be used to achieve spoken-digit recognition with an accuracy similar to that of state-of-the-art neural networks and the regime of magnetization dynamics that leads to the greatest performance is determined.
Neuro-Inspired Computing With Emerging Nonvolatile Memorys
  • Shimeng Yu
  • Computer Science
    Proceedings of the IEEE
  • 2018
TLDR
This comprehensive review summarizes state of the art, challenges, and prospects of the neuro-inspired computing with emerging nonvolatile memory devices and presents a device-circuit-algorithm codesign methodology to evaluate the impact of nonideal device effects on the system-level performance.
HfO2-Based OxRAM Devices as Synapses for Convolutional Neural Networks
TLDR
Electrical characterization results show that the proposed HfO2-based OxRAM technology offers good electrical properties in terms of endurance, speed, and low energy, and thus being well suited for neuromorphic applications.
...
1
2
3
...