Neuromorphic Hardware System for Visual Pattern Recognition With Memristor Array and CMOS Neuron

Chu, Myonglae; Kim, Byoung‐Ho; Park, Sangsu; Hwang, Hyunsang; Jeon, Moongu; Lee, Byoung Hun; Lee, Byung-Geun

doi:10.1109/tie.2014.2356439

Cited by 264 publications

(130 citation statements)

References 23 publications

Supporting

Mentioning

129

Contrasting

Unclassified

Order By: Relevance

“…In addition, these systems make it possible to simulate the properties of synapses and implement solid varieties of artificial neural networks by combining CMOS technology and memristor arrays [4][5][6]. To meet the requirements for portable devices in lowering the switching power consumption and enhancing the high-speed performance, manufacturers are continuously designing micro-and nanoscale MIM structures [1][2][3][4][5][6][7].…”

Section: Introductionmentioning

confidence: 99%

Automatized complex for measuring the electrical properties of MIM structures

Gryaznov

Дорошева

Вохминцев

et al. 2016

2016 International Siberian Conference on Control and Communications (SIBCON)

View full text Add to dashboard Cite

We have designed an automatic complex equipped by a Cascade Microtech MPS150 micro-probing station and a PXIe-4143 source measure unit from National Instruments Corporation to investigate the electrical properties of metal/insulator/metal sandwich structures. We have also succeeded in implementing conditions for the current-voltage characteristics measurements and in simulating procedures for the multiple reading/writing of digital information by means of resistive switching inside the nanostructured layers. We have studied anodized-titanium-dioxide-based Ti/TiO2-NT/Au structures with diameter of 100 micrometers, synthesized by the mask method. For the micromemristors produced, we have estimated the resistance in the low-(<10 kOhm) and high-ohmic (> 50 kOhm) states.

show abstract

Section: Introductionmentioning

confidence: 99%

Automatized complex for measuring the electrical properties of MIM structures

Gryaznov

Дорошева

Вохминцев

et al. 2016

2016 International Siberian Conference on Control and Communications (SIBCON)

View full text Add to dashboard Cite

show abstract

“…Moreover, the AD/DA conversions induce high design complexity as well as significant area and energy overheads. Recently, a spiking-based neuromorphic computing system was demonstrated for pattern recognition [16] and believed to have better tolerance to the signal noise. However, it requires complicated neuron circuits and can run only at an extremely low speed (e.g., ten operations per second).…”

Section: Introductionmentioning

confidence: 99%

A spiking neuromorphic design with resistive crossbar

Liu

Yan

Yang

et al. 2015

Proceedings of the 52nd Annual Design Automation Conference

116

View full text Add to dashboard Cite

Neuromorphic systems recently gained increasing attention for their high computation efficiency. Many designs have been proposed and realized with traditional CMOS technology or emerging devices. In this work, we proposed a spiking neuromorphic design built on resistive crossbar structures and implemented with IBM 130nm technology. Our design adopts a rate coding scheme where pre-and post-neuron signals are represented by digitalized pulses. The weighting function of pre-neuron signals is executed on the resistive crossbar in analog format. The computing result is transferred into digitalized output spikes via an integrate-and-fire circuit (IFC) as the postneuron. We calibrated the computation accuracy of the entire system through circuit simulations. The results demonstrated a good match to our analytic modeling. Furthermore, we implemented both feedforward and Hopfield networks by utilizing the proposed neuromorphic design. The system performance and robustness were studied through massive Monte-Carlo simulations based on the application of digital image recognition. Comparing to the previous crossbar-based computing engine that represents data with voltage amplitude, our design can achieve >50% energy savings, while the average probability of failed recognition increase only 1.46% and 5.99% in the feedforward and Hopfield implementations, respectively.

show abstract

“…It is much simpler compared with neuromorphic system using the memristor as a synaptic device. This is because the memristor is a two-terminal device and requires enormous switching component to receive signals from post-synaptic neuron circuits [8][9][10]. STDP characteristics were emulated with time difference (∆t) between pre-and post-synaptic spikes.…”

Section: Connection With Pre-and Post-synaptic Neuron Circuits Anmentioning

confidence: 99%

“…Recently, there are great interests in realizing artificial synapses based on the Hebbian learning for the brain-like computing architectures [1][2][3][4][5][6][7][8][9][10]. This is because a synapse is believed to make a great contribution to many cognitive functions such as perception and memory in a biological system [11][12][13][14][15][16].…”

Section: Introductionmentioning

confidence: 99%

“…This is because a synapse is believed to make a great contribution to many cognitive functions such as perception and memory in a biological system [11][12][13][14][15][16]. However, one of the strong candidates for artificial synapse, the memristor, is a two terminal device and needs extra selection components to interact with both pre-and post-synaptic neurons circuits [8][9][10]. Moreover, memristors are implemented by the non-silicon devices and have difficulty in being cointegrated with the conventional complementary metaloxide-semiconductor (CMOS) architectures.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Simulation Study on Silicon-Based Floating Body Synaptic Transistor with Short- and Long-Term Memory Functions and Its Spike Timing-Dependent Plasticity

Kim¹,

Cho²,

Sun³

et al. 2016

JSTS:Journal of Semiconductor Technology and Science

View full text Add to dashboard Cite

Abstract-In this work, a novel silicon (Si) based floating body synaptic transistor (SFST) is studied to mimic the transition from short-term memory to long-term one in the biological system. The structure of the proposed SFST is based on an n-type metaloxide-semiconductor field-effect transistor (MOSFET) with floating body and charge storage layer which provide the functions of short-and long-term memories, respectively. It has very similar characteristics with those of the biological memory system in the sense that the transition between shortand long-term memories is performed by the repetitive learning. Spike timing-dependent plasticity (STDP) characteristics are closely investigated for the SFST device. It has been found from the simulation results that the connectivity between pre-and postsynaptic neurons has strong dependence on the relative spike timing among electrical signals. In addition, the neuromorphic system having direct connection between the SFST devices and neuron circuits are designed.Index Terms-Neuromorphic system, synaptic transistor, short-and long-term memory, spike timing-dependent plasticity (STDP)

show abstract

Neuromorphic Hardware System for Visual Pattern Recognition With Memristor Array and CMOS Neuron

Cited by 264 publications

References 23 publications

Automatized complex for measuring the electrical properties of MIM structures

Automatized complex for measuring the electrical properties of MIM structures

A spiking neuromorphic design with resistive crossbar

Simulation Study on Silicon-Based Floating Body Synaptic Transistor with Short- and Long-Term Memory Functions and Its Spike Timing-Dependent Plasticity

Contact Info

Product

Resources

About