• Electronics and Telecommunications Trends
• /
• v.8 no.3
• /
• pp.11-20
• /
• 1993

CMAC 신경회로망은 다차원 비선형 함수를 학습을 통하여 발생되는 많고 복잡한 데이터들을 퍼셉트론과 같이 집합시켜 메모리를 구성하고 처리하는 분야이다. 일반적으로 학습알고리즘은 소수의 반복으로써 수렴한다. 본고에서는 CMAC의 메카니즘 및 CMAC의 특성을 기술하고, CMAC의 학습가능성을 예시하였다. CMAC의 학습성능을 시험하기 위해서 3관절 로봇의 squatting 문제에 적용하였다.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1998.10a
• /
• pp.251-254
• /
• 1998

본 논문에서는 신경회로망의 수렴속도를 높이기 위한 알고리즘을 제안한다. 전형적인 역전파 학습방식은 느린 수렴속도가 단점으로 제기되는데 이는 비용함수의 계곡부근에서 가중치의 궤적이 심한 진동현상을 보이기 때문이다. 이 문제를 해결하기 위해서 본 논문에서는 경사법에서 사용되는 갱신방향을 계곡의 진행방향을 이용하여 변경한다. 모의실험을 통하여 제안된 방법으로 가중치의 궤적에 나타나는 진동을 줄이고 수렴속도를 향상시킬 수 있음을 보인다.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.25 no.1
• /
• pp.91-96
• /
• 2015

In this study, we introduce robust face recognition system with illumination variation realized with the aid of CT preprocessing method. As preprocessing algorithm, Census Transform(CT) algorithm is used to extract locally facial features under unilluminated condition. The dimension reduction of the preprocessed data is carried out by using $(2D)^2$PCA which is the extended type of PCA. Feature data extracted through dimension algorithm is used as the inputs of proposed radial basis function neural networks. The hidden layer of the radial basis function neural networks(RBFNN) is built up by fuzzy c-means(FCM) clustering algorithm and the connection weights of the networks are described as the coefficients of linear polynomial function. The essential design parameters (including the number of inputs and fuzzification coefficient) of the proposed networks are optimized by means of artificial bee colony(ABC) algorithm. This study is experimented with both Yale Face database B and CMU PIE database to evaluate the performance of the proposed system.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2008.05a
• /
• pp.345-348
• /
• 2008

Recently, in Weld Processing field, unmanned and automatic system construction has experienced the rapid growth, and diverse signal processing has been employed in order to translate the exact weld pattern. It this paper, We will suggest the effective neural network which can deride the weld quality in arc weld and monitoring system in real time. In addition, We will present the pre-processing for selecting the study data, and the method to evaluate the wave of weld more precisely and accurately through known Neural Network.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.9 no.3
• /
• pp.246-252
• /
• 2004

This paper presents a position sensorless control system of Switched Reluctance Motor (SRM) using neural network. The control of SRM depends on the commutation of the stator phases in synchronism with the rotor position. The position sensing requirement increases the overall cost and complexity. In this paper, the current-flux-rotor position lookup table based position sensorless operation of SRM is presented. Neural network is used to construct the current-flux-rotor position lookup table, and is trained by sufficient experimental data. Experimental results for a 1-hp SRM is presented for the verification of the proposed sensorless algorithm.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.13 no.2
• /
• pp.407-412
• /
• 2009

Recently, in Weld Processing field, unmanned and automatic system construction has experienced the rapid growth, and diverse signal processing has been employed in order to translate the exact weld pattern. In this paper, We will suggest the effective neural network which can decide the weld quality in arc weld and monitoring system in real time. In addition, We will present the pre-processing for selecting the study data, and the method to evaluate the wave of weld more precisely and accurately through known Neural Network.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.17 no.1
• /
• pp.47-57
• /
• 1992

Subject of this work is the application of neural networks for the signal(motor noise)recognition systems which detects motor failures and employs different signal(noise). Charaoteristics that re-sult from damaghe part and measure of motor construction during working. The four layers neural networks is applied to this examination. And consists of one input layer, two hidden layers, and one output layer, and learns by the back propagation algorithm.The results of this examination show that it the construction and the output power of the testmotor and learning motor are compatible, the damaged part of the testmotor are detected correctly in the system on the other hand, if the motors have different constrcotion but similar output power each other, mislesding results are obtained in this system.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.31 no.3
• /
• pp.50-57
• /
• 2003

In the paper, the design and evaluation of a helicopter trajectory tracking controller are presented. The control algorithm is implemented using the feedback linearization technique and the two time-scale separation architecture. In addition, and on-line adaptive architecture that employs a neural network compensating the model inversion error caused by the deficiency of full knowledge of helicopter dynamic is applied to augment the attitude control system. Trajectory tracking performance of the control system in evaluated using modified TMAN simulation program representing as Apache helicopter. It is show that the on-line neural network in an adaptive control architecture is very effective in dealing with the performance depreciation problem of the trajectory tracking control caused by insufficient information of dynamics.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.11 no.4
• /
• pp.325-330
• /
• 2001

It has been realized that the results of circuit level simulation of neural networks, used for optimization problems, arc much different from those of algorism level simulation. In other words, the outputs converges asymptotically as time elapes, however, the input convergence depends on the value of parasitic conductance connected between input node and ground. Also, this conductance affects system performance. This paper discusses the influence of input conductance on the convergece of the continuous Hopfield neural networks. The convergence has been analyzed for the input and output nodes of neurons. Also, the characteristics of equilibrium points has been analyzed depending on different values of the input conductance.

• Proceedings of the KIEE Conference
• /
• 2011.07a
• /
• pp.1966-1967
• /
• 2011

본 연구에서는 방사형 기저 함수를 이용한 다항식 신경회로망(Polynomial Neural Network) 분류기를 제안한다. 제안된 모델은 PNN을 기본 구조로 하여 1층의 다항식 노드 대신에 다중 출력 형태의 방사형 기저 함수를 사용하여 각 노드가 방사형 기저 함수 신경회로망(RBFNN)을 형성한다. RBFNN의 은닉층에는 fuzzy 클러스터링을 사용하여 입력 데이터의 특성을 고려한 적합도를 사용하였다. 제안된 분류기는 입력변수의 수와 다항식 차수가 모델의 성능을 결정함으로 최적화가 필요하며 본 논문에서는 Multiobjective Particle Swarm Optimization(MoPSO)을 사용하여 모델의 성능뿐만 아니라 모델의 복잡성 및 해석력을 고려하였다. 패턴 분류기로써의 제안된 모델을 평가하기 위해 Iris 데이터를 이용하였다.