Walid Al Misba
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Walid Al Misba has a total of 34 co-authors. Here are the top co-authors by number of shared publications:
- Jayasimha Atulasimha
Virginia Commonwealth University
10 shared publications - Mahadi Rajib
Department of Mechanical and Nuclear Engineering, Virginia Commonwealth University 1 , Richmond, Virginia 23284
3 shared publications - Alexander J. Edwards
Department of Electrical and Computer Engineering, The University of Texas at Dallas, Richardson, TX, USA
2 shared publications Dhritiman Bhattacharya
Department of Mechanical and Nuclear Engineering, Virginia Commonwealth University, Richmond, USA
2 shared publications- Fahim F. Chowdhury
Virginia Commonwealth University
2 shared publications - Joseph S. Friedman
Department of Electrical and Computer Engineering, The University of Texas at Dallas, Richardson, TX, USA
2 shared publications
Latest Publications
- Open Access
- 21 November 2025
- Published by American Chemical Society (ACS) in ACS Applied Materials & Interfaces
- Vol. 17 (48) , 65946-65955
Strain Mediated Voltage Control of Magnetic Anisotropy and Magnetization Reversal in Bismuth-Substituted Yttrium Iron Garnet Films and Mesostructures
We report on magnetic anisotropy modulation in Bismuth-substituted Yttrium Iron Garnet (Bi-YIG) thin films and mesoscale patterned structures deposited on a PMN-PT substrate with the application of a voltage-induced strain. The Bi content is selected for low coercivity and higher magnetostriction than that of YIG, yielding significant changes in the hysteresis loops through the magnetoelastic effect. The piezoelectric substrate is poled along its thickness, which is the [011] direction, by applying a voltage across the PMN-PT/SiO2/Bi-YIG/Pt heterostructure. In situ magneto-optical Kerr effect microscopy (MOKE) shows the modulation of magnetic anisotropy with voltage-induced strain. Furthermore, voltage control of the magnetic domain state of the Bi-YIG film at a fixed magnetic field produces 90° switching of the magnetization easy axis above a threshold voltage. The magnetoelectric coefficient of the heterostructure is 1.05 × 10-7 s m-1 which is competitive with that of other ferromagnetic oxide films on ferroelectric substrates such as La0.67Sr0.33MnO3/PMN-PT and YIG/PMN-PZT. Voltage control of magnetization reversal fields in 5-30 μm wide dots and racetracks of Bi-YIG show potential for energy-efficient nonvolatile memory and neuromorphic computing devices.
- 21 November 2025
- Published by American Chemical Society (ACS) in ACS Applied Materials & Interfaces
- Vol. 17 (48) , 65946-65955
Open AccessStrain Mediated Voltage Control of Magnetic Anisotropy and Magnetization Reversal in Bismuth-Substituted Yttrium Iron Garnet Films and Mesostructures
We report on magnetic anisotropy modulation in Bismuth-substituted Yttrium Iron Garnet (Bi-YIG) thin films and mesoscale patterned structures deposited on a PMN-PT substrate with the application of a voltage-induced strain. The Bi content is selected for low coercivity and higher magnetostriction than that of YIG, yielding significant changes in the hysteresis loops through the magnetoelastic effect. The piezoelectric substrate is poled along its thickness, which is the [011] direction, by applying a voltage across the PMN-PT/SiO2/Bi-YIG/Pt heterostructure. In situ magneto-optical Kerr effect microscopy (MOKE) shows the modulation of magnetic anisotropy with voltage-induced strain. Furthermore, voltage control of the magnetic domain state of the Bi-YIG film at a fixed magnetic field produces 90° switching of the magnetization easy axis above a threshold voltage. The magnetoelectric coefficient of the heterostructure is 1.05 × 10-7 s m-1 which is competitive with that of other ferromagnetic oxide films on ferroelectric substrates such as La0.67Sr0.33MnO3/PMN-PT and YIG/PMN-PZT. Voltage control of magnetization reversal fields in 5-30 μm wide dots and racetracks of Bi-YIG show potential for energy-efficient nonvolatile memory and neuromorphic computing devices.
- Open Access
- 7 July 2025
- Published by IOP Publishing in Nanotechnology
- Vol. 36 (27) , 275202
Quantized artificial neural networks implemented with spintronic stochastic computing
An artificial neural network (ANN) inference involves matrix vector multiplications that require a very large number of multiply and accumulate operations, resulting in high energy cost and large device footprint. Stochastic computing (SC) offers a less resource-intensive ANN implementation with minimal accuracy loss. Random number generators (RNG) are required to implement SC in hardware. These can be realized through stochastic-magnetic tunnel junctions (s-MTJ), where the energy barrier to switch between the ‘up’ and ‘down’ states is designed to be small, enabling thermal noise to generate a random bit stream. While s-MTJs have previously been used to implement SC-ANNs, these studies have been limited to architectures with continuously varying (i.e., analog) weights. In this work, we study the use of SC for matrix vector multiplication with quantized synaptic weights and quantized outputs. We show that a quantized SC-ANN, implemented by using experimentally obtained s-MTJ bitstreams and a limited number of discrete quantized states for both weights and hidden layer nodes in an ANN, can effectively reduce time (latency) and energy consumption in SC compared to an analog implementation, while largely preserving accuracy. We implemented quantization with 5 and 11 quantized states, along with SC configured with stochastic bitstream lengths of 100, 200, 300, 400, and 500 on neural networks with one hidden layer and three hidden layers. Inference was performed on the MNIST dataset for both training with SC and without SC. Training with SC provided better accuracy for all cases. For the shortest bitstream of 100 bits, the highest accuracies were 92% for one hidden layer and over 96% for three hidden layers. The overall system attained its peak accuracy of 96.82% using a 400 bit stochastic bitstream with three hidden layers. Our investigations demonstrate 9× improvement in latency and 2.6× improvement in energy consumption using the quantized SC approach compared to a similar s-MTJ based ANN architecture without quantization.
- 7 July 2025
- Published by IOP Publishing in Nanotechnology
- Vol. 36 (27) , 275202
Open AccessQuantized artificial neural networks implemented with spintronic stochastic computing
An artificial neural network (ANN) inference involves matrix vector multiplications that require a very large number of multiply and accumulate operations, resulting in high energy cost and large device footprint. Stochastic computing (SC) offers a less resource-intensive ANN implementation with minimal accuracy loss. Random number generators (RNG) are required to implement SC in hardware. These can be realized through stochastic-magnetic tunnel junctions (s-MTJ), where the energy barrier to switch between the ‘up’ and ‘down’ states is designed to be small, enabling thermal noise to generate a random bit stream. While s-MTJs have previously been used to implement SC-ANNs, these studies have been limited to architectures with continuously varying (i.e., analog) weights. In this work, we study the use of SC for matrix vector multiplication with quantized synaptic weights and quantized outputs. We show that a quantized SC-ANN, implemented by using experimentally obtained s-MTJ bitstreams and a limited number of discrete quantized states for both weights and hidden layer nodes in an ANN, can effectively reduce time (latency) and energy consumption in SC compared to an analog implementation, while largely preserving accuracy. We implemented quantization with 5 and 11 quantized states, along with SC configured with stochastic bitstream lengths of 100, 200, 300, 400, and 500 on neural networks with one hidden layer and three hidden layers. Inference was performed on the MNIST dataset for both training with SC and without SC. Training with SC provided better accuracy for all cases. For the shortest bitstream of 100 bits, the highest accuracies were 92% for one hidden layer and over 96% for three hidden layers. The overall system attained its peak accuracy of 96.82% using a 400 bit stochastic bitstream with three hidden layers. Our investigations demonstrate 9× improvement in latency and 2.6× improvement in energy consumption using the quantized SC approach compared to a similar s-MTJ based ANN architecture without quantization.
- Open Access
- 10 May 2024
- Published by IOP Publishing in Neuromorphic Computing and Engineering
- Vol. 4 (2) , 024012
Quantized non-volatile nanomagnetic domain wall synapse based autoencoder for efficient unsupervised network anomaly detection
Anomaly detection in real-time using autoencoders implemented on edge devices is exceedingly challenging due to limited hardware, energy, and computational resources. We show that these limitations can be addressed by designing an autoencoder with low-resolution non-volatile memory-based synapses and employing an effective quantized neural network learning algorithm. We further propose nanoscale ferromagnetic racetracks with engineered notches hosting magnetic domain walls (DW) as exemplary non-volatile memory-based autoencoder synapses, where limited state (5-state) synaptic weights are manipulated by spin orbit torque (SOT) current pulses to write different magnetoresistance states. The performance of anomaly detection of the proposed autoencoder model is evaluated on the NSL-KDD dataset. Limited resolution and DW device stochasticity aware training of the autoencoder is performed, which yields comparable anomaly detection performance to the autoencoder having floating-point precision weights. While the limited number of quantized states and the inherent stochastic nature of DW synaptic weights in nanoscale devices are typically known to negatively impact the performance, our hardware-aware training algorithm is shown to leverage these imperfect device characteristics to generate an improvement in anomaly detection accuracy (90.98%) compared to accuracy obtained with floating-point synaptic weights that are extremely memory intensive. Furthermore, our DW-based approach demonstrates a remarkable reduction of at least three orders of magnitude in weight updates during training compared to the floating-point approach, implying significant reduction in operation energy for our method. This work could stimulate the development of extremely energy efficient non-volatile multi-state synapse-based processors that can perform real-time training and inference on the edge with unsupervised data.
- 10 May 2024
- Published by IOP Publishing in Neuromorphic Computing and Engineering
- Vol. 4 (2) , 024012
Open AccessQuantized non-volatile nanomagnetic domain wall synapse based autoencoder for efficient unsupervised network anomaly detection
Anomaly detection in real-time using autoencoders implemented on edge devices is exceedingly challenging due to limited hardware, energy, and computational resources. We show that these limitations can be addressed by designing an autoencoder with low-resolution non-volatile memory-based synapses and employing an effective quantized neural network learning algorithm. We further propose nanoscale ferromagnetic racetracks with engineered notches hosting magnetic domain walls (DW) as exemplary non-volatile memory-based autoencoder synapses, where limited state (5-state) synaptic weights are manipulated by spin orbit torque (SOT) current pulses to write different magnetoresistance states. The performance of anomaly detection of the proposed autoencoder model is evaluated on the NSL-KDD dataset. Limited resolution and DW device stochasticity aware training of the autoencoder is performed, which yields comparable anomaly detection performance to the autoencoder having floating-point precision weights. While the limited number of quantized states and the inherent stochastic nature of DW synaptic weights in nanoscale devices are typically known to negatively impact the performance, our hardware-aware training algorithm is shown to leverage these imperfect device characteristics to generate an improvement in anomaly detection accuracy (90.98%) compared to accuracy obtained with floating-point synaptic weights that are extremely memory intensive. Furthermore, our DW-based approach demonstrates a remarkable reduction of at least three orders of magnitude in weight updates during training compared to the floating-point approach, implying significant reduction in operation energy for our method. This work could stimulate the development of extremely energy efficient non-volatile multi-state synapse-based processors that can perform real-time training and inference on the edge with unsupervised data.
- 12 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks and Learning Systems
- Vol. 36 (3) , 4996-5005
Quantized Magnetic Domain Wall Synapse for Efficient Deep Neural Networks
The quantization of synaptic weights using emerging nonvolatile memory (NVM) devices has emerged as a promising solution to implement computationally efficient neural networks on resource constrained hardware. However, the practical implementation of such synaptic weights is hampered by the imperfect memory characteristics, specifically the availability of limited number of quantized states and the presence of large intrinsic device variation and stochasticity involved in writing the synaptic states. This article presents on-chip training and inference of a neural network using quantized magnetic domain wall (DW)-based synaptic array and CMOS peripheral circuits. A rigorous model of the magnetic DW device considering stochasticity and process variations has been utilized for the synapse. To achieve stable quantized weights, DW pinning has been achieved by means of physical constrictions. Finally, VGG8 architecture for CIFAR-10 image classification has been simulated by using the extracted synaptic device characteristics. The performance in terms of accuracy, energy, latency, and area consumption has been evaluated while considering the process variations and nonidealities in the DW device as well as the peripheral circuits. The proposed quantized neural network (QNN) architecture achieves efficient on-chip learning with 92.4% and 90.4% training and inference accuracy, respectively. In comparison to pure CMOS-based design, it demonstrates an overall improvement in area, energy, and latency by 13.8 $\times$ , 9.6 $\times$ , and 3.5 $\times$ , respectively.
- 12 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks and Learning Systems
- Vol. 36 (3) , 4996-5005
Quantized Magnetic Domain Wall Synapse for Efficient Deep Neural Networks
The quantization of synaptic weights using emerging nonvolatile memory (NVM) devices has emerged as a promising solution to implement computationally efficient neural networks on resource constrained hardware. However, the practical implementation of such synaptic weights is hampered by the imperfect memory characteristics, specifically the availability of limited number of quantized states and the presence of large intrinsic device variation and stochasticity involved in writing the synaptic states. This article presents on-chip training and inference of a neural network using quantized magnetic domain wall (DW)-based synaptic array and CMOS peripheral circuits. A rigorous model of the magnetic DW device considering stochasticity and process variations has been utilized for the synapse. To achieve stable quantized weights, DW pinning has been achieved by means of physical constrictions. Finally, VGG8 architecture for CIFAR-10 image classification has been simulated by using the extracted synaptic device characteristics. The performance in terms of accuracy, energy, latency, and area consumption has been evaluated while considering the process variations and nonidealities in the DW device as well as the peripheral circuits. The proposed quantized neural network (QNN) architecture achieves efficient on-chip learning with 92.4% and 90.4% training and inference accuracy, respectively. In comparison to pure CMOS-based design, it demonstrates an overall improvement in area, energy, and latency by 13.8 $\times$ , 9.6 $\times$ , and 3.5 $\times$ , respectively.
- Open Access
- 7 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 124725-124737
Spintronic Physical Reservoir for Autonomous Prediction and Long-Term Household Energy Load Forecasting
With the growing use of artificial neural networks (ANNs) in temporal data processing tasks, the cost of training for complex ANNs is an escalating concern. Physical reservoir computing (RC), a variation of recurrent neural networks (RNNs), obviate the need for most data intensive matrix vector multiplication in the recurrent layer by evolving the RC’s internal states with the inherent nonlinear dynamics and short-term memory. In this study, we show that magnetic skyrmion confined in a fixed geometry forming the soft layer of a magnetic tunnel junction (MTJ) can work as a RC and perform autonomous long-term prediction of temporal data. Our proposed skyrmion reservoir allows for manipulation of spin dynamics with ultra energy efficient voltage controlled magnetic anisotropy modulation (VCMA) method. Furthermore, the boundary effect on the skyrmion from the geometric edges provides necessary consistency property of the reservoir. We employ our proposed reservoir for the modeling and prediction of the chaotic time series such as Mackey-Glass and dynamic time-series data, such as household building energy loads. For autonomous run, the predicted output is fed to the input of the reservoir. By comparing our spintronic physical RC approach with energy load forecasting algorithms, such as LSTMs and RNNs, we conclude that the proposed framework presents good performance in achieving high predictions accuracy, while also requiring low memory and energy both of which are at a premium in hardware resource and power constrained edge applications. Further, the proposed approach is shown to require very small training datasets and at the same time being at least 16× energy efficient compared to the sequence to sequence LSTM for accurate household load predictions. Higher endurance, fast processing and well-established technology (i.e., MTJ) to integrate with CMOS technology makes such spintronic reservoir attractive over other emergent memory device-based reservoirs.
- 7 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 124725-124737
Open AccessSpintronic Physical Reservoir for Autonomous Prediction and Long-Term Household Energy Load Forecasting
With the growing use of artificial neural networks (ANNs) in temporal data processing tasks, the cost of training for complex ANNs is an escalating concern. Physical reservoir computing (RC), a variation of recurrent neural networks (RNNs), obviate the need for most data intensive matrix vector multiplication in the recurrent layer by evolving the RC’s internal states with the inherent nonlinear dynamics and short-term memory. In this study, we show that magnetic skyrmion confined in a fixed geometry forming the soft layer of a magnetic tunnel junction (MTJ) can work as a RC and perform autonomous long-term prediction of temporal data. Our proposed skyrmion reservoir allows for manipulation of spin dynamics with ultra energy efficient voltage controlled magnetic anisotropy modulation (VCMA) method. Furthermore, the boundary effect on the skyrmion from the geometric edges provides necessary consistency property of the reservoir. We employ our proposed reservoir for the modeling and prediction of the chaotic time series such as Mackey-Glass and dynamic time-series data, such as household building energy loads. For autonomous run, the predicted output is fed to the input of the reservoir. By comparing our spintronic physical RC approach with energy load forecasting algorithms, such as LSTMs and RNNs, we conclude that the proposed framework presents good performance in achieving high predictions accuracy, while also requiring low memory and energy both of which are at a premium in hardware resource and power constrained edge applications. Further, the proposed approach is shown to require very small training datasets and at the same time being at least 16× energy efficient compared to the sequence to sequence LSTM for accurate household load predictions. Higher endurance, fast processing and well-established technology (i.e., MTJ) to integrate with CMOS technology makes such spintronic reservoir attractive over other emergent memory device-based reservoirs.
- Open Access
- 17 August 2023
- Published by Springer Nature in Communications Physics
- Vol. 6 (1) , 1-9
Passive frustrated nanomagnet reservoir computing
Reservoir computing (RC) has received recent interest because reservoir weights do not need to be trained, enabling extremely low-resource consumption implementations, which could have a transformative impact on edge computing and in-situ learning where resources are severely constrained. Ideally, a natural hardware reservoir should be passive, minimal, expressive, and feasible; to date, proposed hardware reservoirs have had difficulty meeting all of these criteria. We, therefore, propose a reservoir that meets all of these criteria by leveraging the passive interactions of dipole-coupled, frustrated nanomagnets. The frustration significantly increases the number of stable reservoir states, enriching reservoir dynamics, and as such these frustrated nanomagnets fulfill all of the criteria for a natural hardware reservoir. We likewise propose a complete frustrated nanomagnet reservoir computing (NMRC) system with low-power complementary metal-oxide semiconductor (CMOS) circuitry to interface with the reservoir, and initial experimental results demonstrate the reservoir’s feasibility. The reservoir is verified with micromagnetic simulations on three separate tasks demonstrating expressivity. The proposed system is compared with a CMOS echo state network (ESN), demonstrating an overall resource decrease by a factor of over 10,000,000, demonstrating that because NMRC is naturally passive and minimal it has the potential to be extremely resource efficient.
- 17 August 2023
- Published by Springer Nature in Communications Physics
- Vol. 6 (1) , 1-9
Open AccessPassive frustrated nanomagnet reservoir computing
Reservoir computing (RC) has received recent interest because reservoir weights do not need to be trained, enabling extremely low-resource consumption implementations, which could have a transformative impact on edge computing and in-situ learning where resources are severely constrained. Ideally, a natural hardware reservoir should be passive, minimal, expressive, and feasible; to date, proposed hardware reservoirs have had difficulty meeting all of these criteria. We, therefore, propose a reservoir that meets all of these criteria by leveraging the passive interactions of dipole-coupled, frustrated nanomagnets. The frustration significantly increases the number of stable reservoir states, enriching reservoir dynamics, and as such these frustrated nanomagnets fulfill all of the criteria for a natural hardware reservoir. We likewise propose a complete frustrated nanomagnet reservoir computing (NMRC) system with low-power complementary metal-oxide semiconductor (CMOS) circuitry to interface with the reservoir, and initial experimental results demonstrate the reservoir’s feasibility. The reservoir is verified with micromagnetic simulations on three separate tasks demonstrating expressivity. The proposed system is compared with a CMOS echo state network (ESN), demonstrating an overall resource decrease by a factor of over 10,000,000, demonstrating that because NMRC is naturally passive and minimal it has the potential to be extremely resource efficient.
- Open Access
- 9 November 2022
- Published by IOP Publishing in Neuromorphic Computing and Engineering
- Vol. 2 (4) , 044011
Skyrmion based energy-efficient straintronic physical reservoir computing
Physical Reservoir Computing (PRC) is an unconventional computing paradigm that exploits the nonlinear dynamics of reservoir blocks to perform temporal data classification and prediction tasks. Here, we show with simulations that patterned thin films hosting skyrmion can implement energy-efficient straintronic reservoir computing (RC) in the presence of room-temperature thermal perturbation. This RC block is based on strain-induced nonlinear breathing dynamics of skyrmions, which are coupled to each other through dipole and spin-wave interaction. The nonlinear and coupled magnetization dynamics were exploited to perform temporal data classification and prediction. Two performance metrics, namely Short-Term Memory (STM) and Parity Check (PC) capacity are studied and shown to be promising (4.39 and 4.62 respectively), in addition to showing it can classify sine and square waves with 100% accuracy. These demonstrate the potential of such skyrmion based PRC. Furthermore, our study shows that nonlinear magnetization dynamics and interaction through spin-wave and dipole coupling have a strong influence on STM and PC capacity, thus explaining the role of physical interaction in a dynamical system on its ability to perform RC.
- 9 November 2022
- Published by IOP Publishing in Neuromorphic Computing and Engineering
- Vol. 2 (4) , 044011
Open AccessSkyrmion based energy-efficient straintronic physical reservoir computing
Physical Reservoir Computing (PRC) is an unconventional computing paradigm that exploits the nonlinear dynamics of reservoir blocks to perform temporal data classification and prediction tasks. Here, we show with simulations that patterned thin films hosting skyrmion can implement energy-efficient straintronic reservoir computing (RC) in the presence of room-temperature thermal perturbation. This RC block is based on strain-induced nonlinear breathing dynamics of skyrmions, which are coupled to each other through dipole and spin-wave interaction. The nonlinear and coupled magnetization dynamics were exploited to perform temporal data classification and prediction. Two performance metrics, namely Short-Term Memory (STM) and Parity Check (PC) capacity are studied and shown to be promising (4.39 and 4.62 respectively), in addition to showing it can classify sine and square waves with 100% accuracy. These demonstrate the potential of such skyrmion based PRC. Furthermore, our study shows that nonlinear magnetization dynamics and interaction through spin-wave and dipole coupling have a strong influence on STM and PC capacity, thus explaining the role of physical interaction in a dynamical system on its ability to perform RC.
- 5 September 2022
- Published by AIP Publishing in Applied Physics Letters
- Vol. 121 (10) , 102402
Focused surface acoustic wave induced nano-oscillator based reservoir computing
We demonstrate using micromagnetic simulations that a nanomagnet array excited by surface acoustic waves (SAWs) can work as a reservoir. An input nanomagnet is excited with focused SAW and coupled to several nanomagnets, seven of which serve as output nanomagnets. To evaluate memory effect and computing capability, we study the short-term memory (STM) and parity check (PC) capacities, respectively. The SAW (4 GHz carrier frequency) amplitude is modulated to provide a sequence of sine and square waves of 100 MHz frequency. The responses of the selected output nanomagnets are processed by reading the envelope of their magnetization states, which is used to train the output weights using the regression method. For classification, a random sequence of 100 square and sine wave samples is used, of which 80% are used for training, and the rest are used for testing. We achieve 100% training and 100% testing accuracy. The average STM and PC are calculated to be ∼4.69 and ∼5.39 bits, respectively, which is indicative of the proposed acoustically driven nanomagnet oscillator array being well suited for physical reservoir computing applications. The energy dissipation is ∼2.5 times lower than a CMOS-based echo-state network. Furthermore, the reservoir is able to accurately predict Mackey-Glass time series up to several time steps ahead. Finally, the ability to use high frequency SAW makes the nanomagnet reservoir scalable to small dimensions, and the ability to modulate the envelope at a lower frequency (100 MHz) adds flexibility to encode different signals beyond the sine/square waves classification and Mackey-Glass predication tasks demonstrated here.
- 5 September 2022
- Published by AIP Publishing in Applied Physics Letters
- Vol. 121 (10) , 102402
Focused surface acoustic wave induced nano-oscillator based reservoir computing
We demonstrate using micromagnetic simulations that a nanomagnet array excited by surface acoustic waves (SAWs) can work as a reservoir. An input nanomagnet is excited with focused SAW and coupled to several nanomagnets, seven of which serve as output nanomagnets. To evaluate memory effect and computing capability, we study the short-term memory (STM) and parity check (PC) capacities, respectively. The SAW (4 GHz carrier frequency) amplitude is modulated to provide a sequence of sine and square waves of 100 MHz frequency. The responses of the selected output nanomagnets are processed by reading the envelope of their magnetization states, which is used to train the output weights using the regression method. For classification, a random sequence of 100 square and sine wave samples is used, of which 80% are used for training, and the rest are used for testing. We achieve 100% training and 100% testing accuracy. The average STM and PC are calculated to be ∼4.69 and ∼5.39 bits, respectively, which is indicative of the proposed acoustically driven nanomagnet oscillator array being well suited for physical reservoir computing applications. The energy dissipation is ∼2.5 times lower than a CMOS-based echo-state network. Furthermore, the reservoir is able to accurately predict Mackey-Glass time series up to several time steps ahead. Finally, the ability to use high frequency SAW makes the nanomagnet reservoir scalable to small dimensions, and the ability to modulate the envelope at a lower frequency (100 MHz) adds flexibility to encode different signals beyond the sine/square waves classification and Mackey-Glass predication tasks demonstrated here.
- Open Access
- 5 August 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 10, 84946-84959
Energy Efficient Learning With Low Resolution Stochastic Domain Wall Synapse for Deep Neural Networks
We demonstrate extremely low resolution quantized (nominally 5-state) synapses with large stochastic variations in synaptic weights can be energy efficient and achieve reasonably high testing accuracies compared to Deep Neural Networks (DNNs) of similar sizes using floating-point precision synaptic weights. Specifically, voltage-controlled domain wall (DW) devices demonstrate stochastic behavior and can only encode limited states; however, they are extremely energy efficient during both training and inference. In this study, we propose both in-situ and ex-situ training algorithms, based on modification of the algorithm proposed by Hubara et al. , 2017 which works well with quantization of synaptic weights, and train several 5-layer DNNs on MNIST dataset using 2-, 3- and 5-state DW devices as a synapse. For in-situ training, a separate high precision memory unit preserves and accumulates the weight gradients which prevents accuracy loss due to weight quantization. For ex-situ training, a precursor DNN is first trained based on weight quantization and DW device model. Moreover, a noise tolerance margin is included in both of the training methods to account for the intrinsic device noise. The highest inference accuracies we obtain after in-situ and ex-situ training are ~ 96.67% and ~96.63%, respectively, which is very close to the baseline accuracy of ~97.1% obtained from a similar topology DNN having floating-point precision weights with no stochasticity. Large inter-state intervals due to quantized weights and noise tolerance margin enables in-situ training with significantly lower number of programming attempts. Our proposed approach demonstrates a possibility of at least two orders of magnitud e energy savings compared to the floating-point approach implemented in CMOS. This approach is specifically attractive for low power intelligent edge devices where the ex-situ learning can be utilized for energy efficient non-adaptive tasks and the in-situ learning can provide the opportunity to adapt and learn in a dynamically evolving environment.
- 5 August 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 10, 84946-84959
Open AccessEnergy Efficient Learning With Low Resolution Stochastic Domain Wall Synapse for Deep Neural Networks
We demonstrate extremely low resolution quantized (nominally 5-state) synapses with large stochastic variations in synaptic weights can be energy efficient and achieve reasonably high testing accuracies compared to Deep Neural Networks (DNNs) of similar sizes using floating-point precision synaptic weights. Specifically, voltage-controlled domain wall (DW) devices demonstrate stochastic behavior and can only encode limited states; however, they are extremely energy efficient during both training and inference. In this study, we propose both in-situ and ex-situ training algorithms, based on modification of the algorithm proposed by Hubara et al. , 2017 which works well with quantization of synaptic weights, and train several 5-layer DNNs on MNIST dataset using 2-, 3- and 5-state DW devices as a synapse. For in-situ training, a separate high precision memory unit preserves and accumulates the weight gradients which prevents accuracy loss due to weight quantization. For ex-situ training, a precursor DNN is first trained based on weight quantization and DW device model. Moreover, a noise tolerance margin is included in both of the training methods to account for the intrinsic device noise. The highest inference accuracies we obtain after in-situ and ex-situ training are ~ 96.67% and ~96.63%, respectively, which is very close to the baseline accuracy of ~97.1% obtained from a similar topology DNN having floating-point precision weights with no stochasticity. Large inter-state intervals due to quantized weights and noise tolerance margin enables in-situ training with significantly lower number of programming attempts. Our proposed approach demonstrates a possibility of at least two orders of magnitud e energy savings compared to the floating-point approach implemented in CMOS. This approach is specifically attractive for low power intelligent edge devices where the ex-situ learning can be utilized for energy efficient non-adaptive tasks and the in-situ learning can provide the opportunity to adapt and learn in a dynamically evolving environment.
- 23 September 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 69 (4) , 1658-1666
Voltage-Controlled Energy-Efficient Domain Wall Synapses With Stochastic Distribution of Quantized Weights in the Presence of Thermal Noise and Edge Roughness
We propose energy-efficient voltage-induced strain control of a domain wall (DW) in a perpendicularly magnetized nanoscale racetrack on a piezoelectric substrate that can implement a multistate synapse to be utilized in neuromorphic computing platforms. Here, strain generated in the piezoelectric is mechanically transferred to the racetrack and modulates the perpendicular magnetic anisotropy (PMA) in a system that has significant interfacial Dzyaloshinskii–Moriya interaction (DMI). When different voltages are applied (i.e., different strains are generated) in conjunction with spin–orbit torque (SOT) due to a fixed current flowing in the heavy metal layer for a fixed time, DWs are translated to different distances and implement different synaptic weights. We have shown using micromagnetic simulations that five-state and three-state synapses can be implemented in a racetrack that is modeled with the inclusion of natural edge roughness and room temperature thermal noise. These simulations show interesting dynamics of DWs due to interaction with roughness-induced pinning sites. Thus, notches need not be fabricated to implement multistate nonvolatile synapses. Such a strain-controlled synapse has an energy consumption of ~1 fJ and could thus be very attractive to implement energy-efficient quantized neural networks, which has been shown recently to achieve near equivalent classification accuracy to the full-precision neural networks.
- 23 September 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 69 (4) , 1658-1666
Voltage-Controlled Energy-Efficient Domain Wall Synapses With Stochastic Distribution of Quantized Weights in the Presence of Thermal Noise and Edge Roughness
We propose energy-efficient voltage-induced strain control of a domain wall (DW) in a perpendicularly magnetized nanoscale racetrack on a piezoelectric substrate that can implement a multistate synapse to be utilized in neuromorphic computing platforms. Here, strain generated in the piezoelectric is mechanically transferred to the racetrack and modulates the perpendicular magnetic anisotropy (PMA) in a system that has significant interfacial Dzyaloshinskii–Moriya interaction (DMI). When different voltages are applied (i.e., different strains are generated) in conjunction with spin–orbit torque (SOT) due to a fixed current flowing in the heavy metal layer for a fixed time, DWs are translated to different distances and implement different synaptic weights. We have shown using micromagnetic simulations that five-state and three-state synapses can be implemented in a racetrack that is modeled with the inclusion of natural edge roughness and room temperature thermal noise. These simulations show interesting dynamics of DWs due to interaction with roughness-induced pinning sites. Thus, notches need not be fabricated to implement multistate nonvolatile synapses. Such a strain-controlled synapse has an energy consumption of ~1 fJ and could thus be very attractive to implement energy-efficient quantized neural networks, which has been shown recently to achieve near equivalent classification accuracy to the full-precision neural networks.