scholarly journals Influence of characteristic variation of oxide semiconductor and comparison of the activation function in neuromorphic hardware

2020 ◽  
Vol 11 (2) ◽  
pp. 232-252
Author(s):  
Hiroya Ikeda ◽  
Hiroki Yamane ◽  
Yuta Takishita ◽  
Mutsumi Kimura ◽  
Yasuhiko Nakashima
Keyword(s):  
Activation Function ◽  
Oxide Semiconductor ◽  
Characteristic Variation ◽  
Neuromorphic Hardware

scholarly journals A spiking and adapting tactile sensor for neuromorphic applications

2020 ◽  
Vol 10 (1) ◽  
Author(s):  
Tom Birkoben ◽  
Henning Winterfeld ◽  
Simon Fichtner ◽  
Adrian Petraru ◽  
Hermann Kohlstedt
Keyword(s):  
Field Effect ◽  
Field Effect Transistor ◽  
Piezoelectric Sensor ◽  
Tactile Sensor ◽  
Neuronal Model ◽  
Oxide Semiconductor ◽  
Analogue Circuit ◽  
Ongoing Research ◽  
Neuromorphic Hardware ◽  
Effect Transistor

Abstract The ongoing research on and development of increasingly intelligent artificial systems propels the need for bio inspired pressure sensitive spiking circuits. Here we present an adapting and spiking tactile sensor, based on a neuronal model and a piezoelectric field-effect transistor (PiezoFET). The piezoelectric sensor device consists of a metal-oxide semiconductor field-effect transistor comprising a piezoelectric aluminium-scandium-nitride (AlxSc1−xN) layer inside of the gate stack. The so augmented device is sensitive to mechanical stress. In combination with an analogue circuit, this sensor unit is capable of encoding the mechanical quantity into a series of spikes with an ongoing adaptation of the output frequency. This allows for a broad application in the context of robotic and neuromorphic systems, since it enables said systems to receive information from the surrounding environment and provide encoded spike trains for neuromorphic hardware. We present numerical and experimental results on this spiking and adapting tactile sensor.

scholarly journals SSTDP: Supervised Spike Timing Dependent Plasticity for Efficient Spiking Neural Network Training

2021 ◽  
Vol 15 ◽  
Author(s):  
Fangxin Liu ◽  
Wenbo Zhao ◽  
Yongbiao Chen ◽  
Zongwu Wang ◽  
Tao Yang ◽  
...  
Keyword(s):  
Neural Networks ◽  
Learning Algorithm ◽  
Activation Function ◽  
Spike Timing ◽  
Extraction Property ◽  
Dependent Plasticity ◽  
Neuromorphic Hardware ◽  
Gradient Based ◽  
Spatio Temporal ◽  
Spike Time Dependent Plasticity

Spiking Neural Networks (SNNs) are a pathway that could potentially empower low-power event-driven neuromorphic hardware due to their spatio-temporal information processing capability and high biological plausibility. Although SNNs are currently more efficient than artificial neural networks (ANNs), they are not as accurate as ANNs. Error backpropagation is the most common method for directly training neural networks, promoting the prosperity of ANNs in various deep learning fields. However, since the signals transmitted in the SNN are non-differentiable discrete binary spike events, the activation function in the form of spikes presents difficulties for the gradient-based optimization algorithms to be directly applied in SNNs, leading to a performance gap (i.e., accuracy and latency) between SNNs and ANNs. This paper introduces a new learning algorithm, called SSTDP, which bridges the gap between backpropagation (BP)-based learning and spike-time-dependent plasticity (STDP)-based learning to train SNNs efficiently. The scheme incorporates the global optimization process from BP and the efficient weight update derived from STDP. It not only avoids the non-differentiable derivation in the BP process but also utilizes the local feature extraction property of STDP. Consequently, our method can lower the possibility of vanishing spikes in BP training and reduce the number of time steps to reduce network latency. In SSTDP, we employ temporal-based coding and use Integrate-and-Fire (IF) neuron as the neuron model to provide considerable computational benefits. Our experiments show the effectiveness of the proposed SSTDP learning algorithm on the SNN by achieving the best classification accuracy 99.3% on the Caltech 101 dataset, 98.1% on the MNIST dataset, and 91.3% on the CIFAR-10 dataset compared to other SNNs trained with other learning methods. It also surpasses the best inference accuracy of the directly trained SNN with 25~32× less inference latency. Moreover, we analyze event-based computations to demonstrate the efficacy of the SNN for inference operation in the spiking domain, and SSTDP methods can achieve 1.3~37.7× fewer addition operations per inference. The code is available at: https://github.com/MXHX7199/SNN-SSTDP.

Robust and energy-efficient expression recognition based on improved deep ResNets

2019 ◽  
Vol 64 (5) ◽  
pp. 519-528
Author(s):  
Yunhua Chen ◽  
Jin Du ◽  
Qian Liu ◽  
Ling Zhang ◽  
Yanjun Zeng
Keyword(s):  
Neural Networks ◽  
Energy Consumption ◽  
Facial Expression ◽  
Energy Efficient ◽  
Facial Expression Recognition ◽  
Activation Function ◽  
Low Energy ◽  
Expression Recognition ◽  
Severe Problem ◽  
Neuromorphic Hardware

Abstract To improve the robustness and to reduce the energy consumption of facial expression recognition, this study proposed a facial expression recognition method based on improved deep residual networks (ResNets). Residual learning has solved the degradation problem of deep Convolutional Neural Networks (CNNs); therefore, in theory, a ResNet can consist of infinite number of neural layers. On the one hand, ResNets benefit from better performance on artificial intelligence (AI) tasks, thanks to its deeper network structure; meanwhile, on the other hand, it faces a severe problem of energy consumption, especially on mobile devices. Hence, this study employs a novel activation function, the Noisy Softplus (NSP), to replace rectified linear units (ReLU) to get improved ResNets. NSP is a biologically plausible activation function, which was first proposed in training Spiking Neural Networks (SNNs); thus, NSP-trained models can be directly implemented on ultra-low-power neuromorphic hardware. We built an 18-layered ResNet using NSP to perform facial expression recognition across datasets Cohn-Kanade (CK+), Karolinska Directed Emotional Faces (KDEF) and GENKI-4K. The results achieved better anti-noise ability than ResNet using the activation function ReLU and showed low energy consumption running on neuromorphic hardware. This study not only contributes a solution for robust facial expression recognition, but also consolidates the low energy cost of their implementation on neuromorphic devices, which could pave the way for high-performance, noise-robust and energy-efficient vision applications on mobile hardware.

scholarly journals Artificial 2D van der Waals Synapse Devices via Interfacial Engineering for Neuromorphic Systems

Nanomaterials ◽  
2020 ◽  
Vol 10 (1) ◽  
pp. 88 ◽  
Author(s):  
Woojin Park ◽  
Hye Yeon Jang ◽  
Jae Hyeon Nam ◽  
Jung-Dae Kwon ◽  
Byungjin Cho ◽  
...  
Keyword(s):  
Complementary Metal Oxide Semiconductor ◽  
Metal Electrode ◽  
Pulse Mode ◽  
Oxide Semiconductor ◽  
Interfacial Engineering ◽  
Hardware System ◽  
Artificial Synapse ◽  
Neuromorphic Hardware ◽  
Neuromorphic Systems ◽  
Silicon Based

Despite extensive investigations of a wide variety of artificial synapse devices aimed at realizing a neuromorphic hardware system, the identification of a physical parameter that modulates synaptic plasticity is still required. In this context, a novel two-dimensional architecture consisting of a NbSe2/WSe2/Nb2O5 heterostructure placed on an SiO2/p+ Si substrate was designed to overcome the limitations of the conventional silicon-based complementary metal-oxide semiconductor technology. NbSe2, WSe2, and Nb2O5 were used as the metal electrode, active channel, and conductance-modulating layer, respectively. Interestingly, it was found that the post-synaptic current was successfully modulated by the thickness of the interlayer Nb2O5, with a thicker interlayer inducing a higher synapse spike current and a stronger interaction in the sequential pulse mode. Introduction of the Nb2O5 interlayer can facilitate the realization of reliable and controllable synaptic devices for brain-inspired integrated neuromorphic systems.

scholarly journals Cointegration of single-transistor neurons and synapses by nanoscale CMOS fabrication for highly scalable neuromorphic hardware

2021 ◽  
Vol 7 (32) ◽  
pp. eabg8836
Author(s):  
Joon-Kyu Han ◽  
Jungyeop Oh ◽  
Gyeong-Jun Yun ◽  
Dongeun Yoo ◽  
Myung-Su Kim ◽  
...  
Keyword(s):  
Metal Oxide ◽  
Complementary Metal Oxide Semiconductor ◽  
Metal Oxide Semiconductor ◽  
Oxide Semiconductor ◽  
Nanoscale Cmos ◽  
Neuronal Inhibition ◽  
Neuromorphic Hardware ◽  
Cmos Compatibility ◽  
Effect Transistor ◽  
Letter Pattern

Cointegration of multistate single-transistor neurons and synapses was demonstrated for highly scalable neuromorphic hardware, using nanoscale complementary metal-oxide semiconductor (CMOS) fabrication. The neurons and synapses were integrated on the same plane with the same process because they have the same structure of a metal-oxide semiconductor field-effect transistor with different functions such as homotype. By virtue of 100% CMOS compatibility, it was also realized to cointegrate the neurons and synapses with additional CMOS circuits. Such cointegration can enhance packing density, reduce chip cost, and simplify fabrication procedures. The multistate single-transistor neuron that can control neuronal inhibition and the firing threshold voltage was achieved for an energy-efficient and reliable neural network. Spatiotemporal neuronal functionalities are demonstrated with fabricated single-transistor neurons and synapses. Image processing for letter pattern recognition and face image recognition is performed using experimental-based neuromorphic simulation.

scholarly journals Multi-Terminal Memristive Devices Enabling Tunable Synaptic Plasticity in Neuromorphic Hardware: A Mini-Review

2021 ◽  
Vol 3 ◽  
Author(s):  
Yann Beilliard ◽  
Fabien Alibart
Keyword(s):  
Synaptic Plasticity ◽  
Computational Neuroscience ◽  
Complementary Metal Oxide Semiconductor ◽  
Oxide Semiconductor ◽  
Network Activity ◽  
Neuromorphic Hardware ◽  
Learning Capabilities ◽  
On Line ◽  
Heterosynaptic Plasticity ◽  
On Line Learning

Neuromorphic computing based on spiking neural networks has the potential to significantly improve on-line learning capabilities and energy efficiency of artificial intelligence, specially for edge computing. Recent progress in computational neuroscience have demonstrated the importance of heterosynaptic plasticity for network activity regulation and memorization. Implementing heterosynaptic plasticity in hardware is thus highly desirable, but important materials and engineering challenges remain, calling for breakthroughs in neuromorphic devices. In this mini-review, we propose an overview of the latest advances in multi-terminal memristive devices on silicon with tunable synaptic plasticity, enabling heterosynaptic plasticity in hardware. The scalability and compatibility of the devices with industrial complementary metal oxide semiconductor (CMOS) technologies are discussed.

TEM study of Cosi2 formation via annealing of Co-Ti bilayers on Si

1995 ◽  
Vol 53 ◽  
pp. 464-465
Author(s):  
N. David Theodore ◽  
Andre Vantomme ◽  
Peter Crazier
Keyword(s):  
Thermal Instability ◽  
Lattice Mismatch ◽  
Field Effect Transistors ◽  
Contact Layer ◽  
Interface Roughness ◽  
Oxide Semiconductor ◽  
Surface Layers ◽  
Silicide Layer ◽  
Small Lattice ◽  
Buried Layers

Contact is typically made to source/drain regions of metal-oxide-semiconductor field-effect transistors (MOSFETs) by use of TiSi2 or CoSi2 layers followed by AI(Cu) metal lines. A silicide layer is used to reduce contact resistance. TiSi2 or CoSi2 are chosen for the contact layer because these silicides have low resistivities (~12-15 μΩ-cm for TiSi2 in the C54 phase, and ~10-15 μΩ-cm for CoSi2). CoSi2 has other desirable properties, such as being thermally stable up to >1000°C for surface layers and >1100°C for buried layers, and having a small lattice mismatch with silicon, -1.2% at room temperature. During CoSi2 growth, Co is the diffusing species. Electrode shorts and voids which can arise if Si is the diffusing species are therefore avoided. However, problems can arise due to silicide-Si interface roughness (leading to nonuniformity in film resistance) and thermal instability of the resistance upon further high temperature annealing. These problems can be avoided if the CoSi2 can be grown epitaxially on silicon.

Extranuclear effects of estrogen on cortical bone in males is dependent on estrogen receptor A activation function-1

2016 ◽  
Author(s):  
Helen Farman ◽  
Jianyao Wu ◽  
Karin Gustafsson ◽  
Sara Windahl ◽  
Sung Kim ◽  
...  
Keyword(s):  
Estrogen Receptor ◽  
Cortical Bone ◽  
Activation Function

SCORING MODELING BASED ON NEURAL NETWORKS FOR DETERMINING A BANK BORROWER'S RATING

2020 ◽  
Vol 2020 (10) ◽  
pp. 54-62
Author(s):  
Oleksii VASYLIEV ◽  
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Network Architecture ◽  
Statistical Data ◽  
Activation Function ◽  
Decision Making Process ◽  
Neural Network Architecture ◽  
Acceptable Accuracy ◽  
The Neural Network ◽  
Sigmoid Activation Function

The problem of applying neural networks to calculate ratings used in banking in the decision-making process on granting or not granting loans to borrowers is considered. The task is to determine the rating function of the borrower based on a set of statistical data on the effectiveness of loans provided by the bank. When constructing a regression model to calculate the rating function, it is necessary to know its general form. If so, the task is to calculate the parameters that are included in the expression for the rating function. In contrast to this approach, in the case of using neural networks, there is no need to specify the general form for the rating function. Instead, certain neural network architecture is chosen and parameters are calculated for it on the basis of statistical data. Importantly, the same neural network architecture can be used to process different sets of statistical data. The disadvantages of using neural networks include the need to calculate a large number of parameters. There is also no universal algorithm that would determine the optimal neural network architecture. As an example of the use of neural networks to determine the borrower's rating, a model system is considered, in which the borrower's rating is determined by a known non-analytical rating function. A neural network with two inner layers, which contain, respectively, three and two neurons and have a sigmoid activation function, is used for modeling. It is shown that the use of the neural network allows restoring the borrower's rating function with quite acceptable accuracy.

PENELITIAN DAN PENGEMBANGAN SENSOR AROMATIK SEBAGAI ALAT DETEKSI ALKOHOL PADA PARFUM DAN WEWANGIAN UNTUK AUTENTIKASI PRODUK PARFUM HALAL

2017 ◽  
Vol 18 (2) ◽  
pp. 302-322
Author(s):  
Fajar Hardoyono
Keyword(s):  
Sensor Array ◽  
Principal Component ◽  
Metal Oxide Semiconductor ◽  
Oxide Semiconductor ◽  
Response Patterns ◽  
Data Recording ◽  
Analysis Method ◽  
Sample Group ◽  
Array Pattern ◽  
Group Data

Abstract: The development of aromatic sensor array instrument for the detection of alcohol in perfume. The research was conducted by developing the sensor array using 8 sensors made of metal oxide semiconductor. The sensor types used in this study consisted of TGS 813, TGS 822, TGS 2600, TGS 826, TGS 2611, TGS 2620, TGS 2612 and TGS 2602. Response patterns of 8 sensors formed a sensor array pattern used to detect the aroma of 2 groups of samples perfume made from the essential oil of ginger. The first sample group is pure ginger atsiri oil without mixed alcohol. The second sample group was made from the ginger atsiri oil mixed with alcohol with a level of 0.02 M. The results of the data recording show that the developed instrument is able to dissect the first sample group with the second sample group. Data analysis using principal component analysis method (PCA shows that the instrument is able to distinguish the contaminated alcohol perfume group 0.2 M with the alcohol-free perfume group with 100% accuracy. Keywords: Sensor Aroma, Perfume.

Sign in / Sign up

Export Citation Format

Share Document