• 전자공학회논문지C
• /
• 제35C권2호
• /
• pp.71-79
• /
• 1998

In this paper, we propose a new mART(modified ART) neural network by combining the winner neuron definition method of SOM(self-organizing map) and the real-time adaptive clustering function of ART(adaptive resonance theory) and construct it in a modular structure, for the purpose of organizing the feature maps of three dimensional targets. Being constructed in a modular structure, the proposed modular mART can effectively prevent the clusters from representing multiple classes and can be trained to organze two dimensional distortion invariant feature maps so as to recognize targets with three dimensional distortion. We also present the recognition result and self-organization perfdormance of the proposed modular mART neural network after carried out some experiments with 14 tank and fighter target models.

• ETRI Journal
• /
• 제35권2호
• /
• pp.207-217
• /
• 2013

This paper introduces the modular cellular neural network (CNN), which is a new CNN structure constructed from nine one-layer modules with intercellular interactions between different modules. The new network is suitable for implementing many image processing operations. Inputting an image into the modules results in nine outputs. The topographic characteristic of the cell interactions allows the outputs to introduce new properties for image processing tasks. The stability of the system is proven and the performance is evaluated in several image processing applications. Experiment results on texture segmentation show the power of the proposed structure. The performance of the structure in a real edge detection application using the Berkeley dataset BSDS300 is also evaluated.

• 한국통신학회논문지
• /
• 제29권5C호
• /
• pp.618-626
• /
• 2004

본 논문에서는 로봇의 endeffector를 인식하기 위하여 모듈라 신경회로망인식 시스템을 제안 및 구현하였다. 본 논문에서 제안한 로봇 endeffector 인식시스템은 영상을 획득하고 획득한 영상에서 전처리를 이용하여 로봇의 enddffector를 검색하기 위한 특징 값들을 구한다. 3차원 공간에서 로봇의 endeffector는 움직임에 따라 다양한 형태로 변화하므로 빠르고 정확하게 endeffector를 인식하기 위하여 위치검출 신경회로망 모듈과 크기 검출 신경회로망 모듈로 이루어진 다중모듈신경회로망(MNN; Modular Neural Network)을 이용한다. 이렇게 함으로써 각각의 모듈들에 신경회로망의 인식 능력을 이용하여 로봇 endeffector를 인식하고 좀더 빠른 시간 내에 위치 및 크기를 검출하도록 하는 로봇 endeffector 인식시스템을 구성하도록 하였다. 본 논문에서 제안한 인식 시스템은 잡음에 덜 민감하며 로봇의 endfeector를 인식하는데 좋은 성능을 보임을 알 수 있다. 다중 모듈 신경회로망을 이용한 방법은 기존의 단일 신경회로망보다 14％ 향상된 94％의 인식률을 보이며 원격지에 있는 운영자의 편의를 위해 로봇의 endeffector를 인식하여 화면의 정 중앙에 정확히 위치시킬 수 있다.

• 제어로봇시스템학회논문지
• /
• 제5권5호
• /
• pp.550-557
• /
• 1999

This paper describes a road-following controller using the proposed neural network for autonomous vehicle. Road-following with visual sensor like camera requires intelligent control algorithm because analysis of relation from road image to steering control is complex. The proposed neural network, relative similarity modular network(RSMN), is composed of some learning networks and a partitioniing network. The partitioning network divides input space into multiple sections by similarity of input data. Because divided section has simlar input patterns, RSMN can learn nonlinear relation such as road-following with visual control easily. Visual control uses two criteria on road image from camera; one is position of vanishing point of road, the other is slope of vanishing line of road. The controller using neural network has input of two criteria and output of steering angle. To confirm performance of the proposed neural network controller, a software is developed to simulate vehicle dynamics, camera image generation, visual control, and road-following. Also, prototype autonomous electric vehicle is developed, and usefulness of the controller is verified by physical driving test.

• 한국지반공학회논문집
• /
• 제16권2호
• /
• pp.71-77
• /
• 2000

In this paper, consolidation settlements with time at vertical drain sites were predicted by artificial neural networks. Laboratory test results and field measurements of two vertical drain sites were used for training and testing neural networks. Predicted consolidation settlements by trained artificial neural networks were compared with measured settlements by field instrumentation. To improve the prediction accuracy, modular artificial neural networks were studied. From the results of applying artificial neural networks to the same situation, it was shown that modular artificial neural network model was more accurate for the prediction of the consolidation settlements than the general model.

• 대한음성학회지:말소리
• /
• 제48호
• /
• pp.93-105
• /
• 2003

Since telephone channel has bandlimited frequency characteristics, speech signal over the telephone channel shows degraded speech quality. In this paper, we propose an algorithm using neural network to reconstruct wideband speech from its narrowband version. Although single neural network is a good tool for direct mapping, it has difficulty in training for vast and complicated data. To alleviate this problem, we modularize the neural networks based on appropriate clustering of the acoustic space. We also introduce fuzzy computing to compensate for probable misclassification at the cluster boundaries. According to our simulation, the proposed algorithm showed improved performance over the single neural network and conventional codebook mapping method in both objective and subjective evaluations.

• 한국정보통신학회:학술대회논문집
• /
• 한국해양정보통신학회 2001년도 춘계종합학술대회
• /
• pp.369-373
• /
• 2001

In this paper, we proposes the steps adopted to construct the neural network model for GMAW welds. Conventional, automated process generally involves sophisticated sensing and control techniques applied to various processing parameters. Welding parameters are influenced by numerous factors, such as welding current, arc voltage, torch travel speed, electrode condition and shielding gas type and flow rate etc. In traditional work, the structural mathematical models have been used to represent this relationship. Contrary to the traditional model method, neural network models are based on non-parametric modeling techniques. For the welding process modeling, the non-linearity at well as the coupled input characteristics makes it apparent that the neural network is probably the most suitable candidate for this task. Finally, a suitable proposal to improve the construction of the model has also been presented in the paper.

• 한국전기전자재료학회:학술대회논문집
• /
• 한국전기전자재료학회 2008년도 춘계학술대회 논문집 센서 박막재료연구회 및 광주 전남지부
• /
• pp.54-57
• /
• 2008

A new modular Lateral Information Propagation Networks can be implemented in a IC chip with the circuit VLSI technology for detection of bad materials. The proposed modular architecture is propagated the neural network through inter module connections. For such inter module connections, the host(computer or logic) mediates the exchange of information among modules. Also border nodes in each module have capacitors for temporarily retaining the information from outer modules. For detecting of Faulty Insulator, $4\;{\times}\;4$ neural network modules has been designed and simulation of interpolation with the designed networks has been done.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2002년도 하계종합학술대회 논문집(3)
• /
• pp.139-142
• /
• 2002

The brain of the human is the best model for the artificial intelligence and is studied by many natural, medical scientists and engineers. In the engineering department, the brain model becomes a main subject in the area of development of a system that can represent and think like human. In this paper, we approach and define the function of the brain biologically and especially, make a model for the function of cerebral cortex, known as a part that performs behavior inference and decision for sensitive information from the thalamus. Therefore, we try to make a model for the transfer process of the brain. The brain takes the sensory information from sensory organ, proceeds behavior inference and decision and finally, commands behavior to the motor nerves. We use the modular neural network in this model. finally, we would like to design the intelligent system that can sense, recognize, think and decide like the brain by learning the information process in the brain with the modular neural network.

• 한국컴퓨터정보학회논문지
• /
• 제4권4호
• /
• pp.86-93
• /
• 1999

본 논문에서는 다중 모듈러 신경회로망과 보상입력 알고리즘을 제안하였다. 전자는 신경회로망의 고질적인 문제중의 하나인 수렴속도의 감소를 위하여 제안하였고, 후자는 신경회로망의 인식수행능력 향상을 도모하기 위하여 제안하였다. 본 논문의 실험구성은 두 가지 형태와 시뮬레이션으로 나누어 구성하였다. 첫째로 다중 신경회로망의 구조에 한글, 영문자 와 숫자를 적용하여 인식 실험하였다. 둘째로, 보상입력 알고리즘과 보상입력을 결정하는 단계를 기술하였다. 제안된 알고리즘을 한글, 영문자. 숫자인식에 적용하여 기존의 신경회로망과 비교 평가하였다. 실험결과. 본 논문에서 제안된 모듈러 신경회로망이 기존의 신경회로망에 비하여 3배 이상 수렴속도가 개선되었고 보정입력 알고리즘을 적용한 다중 모듈러 신경회로망은 기존의 신경회로망에 비하여 10%정도 인식률이 향상됨을 고찰하였다.