• JSTS:Journal of Semiconductor Technology and Science
• /
• 제2권1호
• /
• pp.70-81
• /
• 2002

This paper describes a new basic element which shows a synaptic operation for neural logic applications and shows function feasibility. A key device for the logic operation is the insulated-gate pn-junction device on SOI substrates. The basic element allows an interface quite compatible to that of conventional CMOS circuits and vMOS circuits.

• 세라미스트
• /
• 제21권2호
• /
• pp.4-18
• /
• 2018

Lastly, neuromorphic computing chip has been extensively studied as the technology that directly mimics efficient calculation algorithm of human brain, enabling a next-generation intelligent hardware system with high speed and low power consumption. Three-terminal based synaptic transistor has relatively low integration density compared to the two-terminal type memristor, while its power consumption can be realized as being so low and its spike plasticity from synapse can be reliably implemented. Also, the strong electrical interaction between two or more synaptic spikes offers the advantage of more precise control of synaptic weights. In this review paper, the results of synaptic transistor mimicking synaptic behavior of the brain are classified according to the channel material, in order of silicon, organic semiconductor, oxide semiconductor, 1D CNT(carbon nanotube) and 2D van der Waals atomic layer present. At the same time, key technologies related to dielectrics and electrolytes introduced to express hysteresis and plasticity are discussed. In addition, we compared the essential electrical characteristics (EPSC, IPSC, PPF, STM, LTM, and STDP) required to implement synaptic transistors in common and the power consumption required for unit synapse operation. Generally, synaptic devices should be integrated with other peripheral circuits such as neurons. Demonstration of this neuromorphic system level needs the linearity of synapse resistance change, the symmetry between potentiation and depression, and multi-level resistance states. Finally, in order to be used as a practical neuromorphic applications, the long-term stability and reliability of the synapse device have to be essentially secured through the retention and the endurance cycling test related to the long-term memory characteristics.

• Journal of Semiconductor Engineering
• /
• 제2권3호
• /
• pp.142-153
• /
• 2021

Recent research on synaptic devices has been reviewed from the perspective of hardware-based neuromorphic computing. In addition, the backgrounds of neuromorphic computing and two training methods for hardware-based neuromorphic computing are described in detail. Moreover, two types of memristor- and CMOS-based synaptic devices were compared in terms of both the required performance metrics and low-power applications. Based on a review of recent studies, additional power-scalable synaptic devices such as tunnel field-effect transistors are suggested for a plausible candidate for neuromorphic applications.

• JSTS:Journal of Semiconductor Technology and Science
• /
• 제17권2호
• /
• pp.210-215
• /
• 2017

We have developed the neuromorphic system that can work with the four-terminal Si-based synaptic devices and verified the operation of the system using simulation tool and printed-circuit-board (PCB). The symmetrical current mirrors connected to the n-channel and p-channel synaptic devices constitute the synaptic integration part to express the excitation and the inhibition mechanism of neurons, respectively. The number and the weight of the synaptic devices affect the amount of the current reproduced from the current mirror. The double-stage inverters controlling delay time and the NMOS with large threshold voltage ($V_T$) constitute the action-potential generation part. The generated action-potential is transmitted to next neuron and simultaneously returned to the back gate of the synaptic device for changing its weight based on spike-timing-dependent-plasticity (STDP).

• 한국정밀공학회:학술대회논문집
• /
• 한국정밀공학회 1995년도 춘계학술대회 논문집
• /
• pp.273-277
• /
• 1995

Many hudraulic components have nonlinearities to some extent. These nonlinearities often cause the time delay, thus degrading the performance of the hydraulic control systems and making it difficult to modelthem. In this paper, a new vibration isolation control algorithm that eliminates the necessity of a sophiscated modeling of hydraulic system was proposed. The algotithm is a hybrid type control shecheme consisting of a linear controller and a hetero-synaptic neural network controller. Using this control scheme, simulations and experiments were performed for 1 DOF(Degree of freedom) and 2 DOF vibration isolation. The hybrid type control algorithm can isolate the base vibration signifcantly rather than linear control algorithm. And from the weights in hetero-synaptic neural network, we can get the 2nd equivalent differentialmodel of the hydraulic control system with on-line control operation. This equivalent model provides us with much information, such as stability and the characteristics of the control system.

• 한국지능시스템학회:학술대회논문집
• /
• 한국퍼지및지능시스템학회 1993년도 Fifth International Fuzzy Systems Association World Congress 93
• /
• pp.838-841
• /
• 1993

This paper describes a fuzzy network circuit of analogue and digital mixed operation. The circuits are suggested for membership function, MIN function and normalization function using either linear voltage-controlled MOSFET resistance or pulse stream operation. The analogue-digital hybrid fuzzy hardware is extensible to the fuzzy-neural network as its basic configurations are already used in URAN-I of 135,424 synaptic connections.

• 한국지능시스템학회논문지
• /
• 제24권5호
• /
• pp.457-461
• /
• 2014

생물학적 신경 세포의 모델링을 위한 펄스타입 실리콘 뉴런 회로를 $0.18{\mu}m$ CMOS 공정을 이용하여 반도체 집적회로로 설계하였다. 제안하는 뉴런 회로는 입력 전류신호를 위한 커패시터 입력단과, 출력 전압신호 생성을 위한 증폭단 및 펄스신호 초기화를 위한 MOS 스위치로 구성된다. 전압신호 입력을 전류신호 출력으로 변환하는 기능의 시냅스 회로는 몇 개의 PMOS와 NMOS 트랜지스터로 이루어지는 범프회로를 사용한다. 제안하는 뉴런 모델의 검증을 위하여, 2개의 뉴런과 시냅스가 직렬연결된 뉴런체인을 구성하여 SPICE 모의실험을 실시하였다. 모의실험 결과, 뉴런신호의 생성과 시냅스 전달특성의 정상적인 동작을 확인하였다.

• Journal of information and communication convergence engineering
• /
• 제1권3호
• /
• pp.104-108
• /
• 2003

A biologically inspired fuzzy multilayer perceptron is proposed in this paper. The proposed algorithm is established under consideration of biological neuronal structure as well as fuzzy logic operation. We applied this suggested learning algorithm to benchmark problem in neural network such as exclusive OR and 3-bit parity, and to digit image recognition problems. For the comparison between the existing and proposed neural networks, the convergence speed is measured. The result of our simulation indicates that the convergence speed of the proposed learning algorithm is much faster than that of conventional backpropagation algorithm. Furthermore, in the image recognition task, the recognition rate of our learning algorithm is higher than of conventional backpropagation algorithm.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1991년도 추계학술대회 논문집 학회본부
• /
• pp.189-192
• /
• 1991

This paper presents a modular structure design of analog Hopfield neural network. Each multiplier consists of four MOS transistors which are connected to an op-amp at the front end of a neuron. A pair of MOS transistor is used in order to maintain linear operation of the synapse and can produce positive or negative synaptic weight. This architecture can be expandable to any size neural network by forming tree structure. By altering the connections, other nework paradigms can also be implemented using this basic modules. The stength of this approach is the expandability and the general applicability. The layout design of a four-neuron fully connected feedback neural network is presented and is simulated using SPICE. The network shows correct retrival of distorted patterns.

• 전기전자학회논문지
• /
• 제26권4호
• /
• pp.633-638
• /
• 2022

최근 인간의 뇌를 모방한 스파이킹 뉴럴 네트워크(SNNs)의 뉴로모픽(Neuromorphic) 시스템이 주목을 받고 있다. 뉴로모픽 기술은 인지 응용과 처리 과정에서 속도가 빠르고 전력 소모가 적다는 장점이 있다. SNNs 기반의 저항성 랜덤 엑세스 메모리(RRAM) 은 병렬 연산을 위한 가장 효율적인 구조이며 스파이크 타이밍 종속 가소성(STDP)의 점진적인 스위칭 동작을 수행한다. 시냅스 소자 동작으로서의 RRAM은 저 전력 프로세싱과 다양한 메모리 상태를 표현한다. 하지만, RRAM 소자의 통합은 높은 스위칭 전압 및 전류를 유발하여 높은 전력 소비를 초래한다. RRAM의 동작 전압을 낮추기 위해서는 스위칭 레이어와 금속 전극의 신소재를 개발하는 것이 중요하다. 본 연구에서는 스위칭 전압을 낮추기 위해 전기적, 기계적 특성이 우수한 단일 벽 탄소나노튜브(SWCNTs)를 갖는 (Metal/Al₂O₃/HfO_x/SWCNTs/N+silicon, MOCS)라는 최적화된 새로운 구조를 제안하였다. 따라서 SWCNTs 기반 멤리스터의 점진적인 스위칭 동작 및 저 전력 I/V 곡선의 향상을 보여준다.