• Journal of the Korean Operations Research and Management Science Society
• /
• v.20 no.3
• /
• pp.17-29
• /
• 1995

Recently, neural network models have been employed as an alternative to regression analysis for point estimation or function fitting in various field. Thus far, however, no theoretical or empirical guides seem to exist for selecting the tool which the most suitable one for a specific function-fitting problem. In this paper, we evaluate performance of three major function-fitting techniques, regression analysis and two neural network models, back-propagation and linear-Hebbian-learning neural networks. The functions to be fitted are simple linear ones of a single independent variable. The factors considered are size of noise both in dependent and independent variables, portion of outliers, and size of the data. Based on comutational results performed in this study, some guidelines are suggested to choose the best technique that can be used for a specific problem concerned.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.14 no.1
• /
• pp.93-99
• /
• 2004

Neural oscillator can be applied to oscillatory systems such as analyses of image information, voice recognition and etc. Conventional EBPA (Error back Propagation Algorithm) is not proper for oscillatory systems with the complicate input`s patterns because of its tedious training procedures and sluggish convergence problems. However, these problems can be easily solved by using a synchrony characteristic of neural oscillator with PLL(Phase Locked Loop) function and by using a simple Hebbian learning rule. Therefore, in this paper, a technique for Recognition of the Korean Alphabet using Phase Synchronized Neural Oscillator was introduced.

• The KIPS Transactions:PartB
• /
• v.11B no.4
• /
• pp.449-456
• /
• 2004

This paper proposes a novel recognition model, a recurrent fuzzy associative memory(RFAM), for recognizing time-series patterns contained an ambiguity. RFAM is basically extended from FAM(Fuzzy Associative memory) by adding a recurrent layer which can be used to deal with sequential input patterns and to characterize their temporal relations. RFAM provides a Hebbian-style learning method which establishes the degree of association between input and output. The error back-propagation algorithm is also adopted to train the weights of the recurrent layer of RFAM. To evaluate the performance of the proposed model, we applied it to a word boundary detection problem of speech signal.

• 제어로봇시스템학회:학술대회논문집
• /
• 2001.10a
• /
• pp.115.3-115
• /
• 2001

In this paper, discrete-time cellular neural networks are designed in order to function as associative memories by using Hebbian learning rule and non-cloning template. The proposed method has a very simple structure to design and to learn. Weights are updated by the connection between the neuron and its neighborhood. In the simulation, the proposed method is applied to the classification of a traffic sign pattern.

• Proceedings of the Korean Information Science Society Conference
• /
• 1998.10c
• /
• pp.333-335
• /
• 1998

에러 역전파 신경망에서 학습속도와 수렴률은 초기 가중의 분포에 따라 크게 영향을 받는다. 본 연구에서는 이를 위하여 비교사 학습 신경망(Hebbian learning rule)을 이용한 새로운 초기 가중치 결정 방법을 제안한다. 또는 비교사 학습 신경망이 에러 역전파 신경망 학습에 적당하도록 은닉층의 각 뉴런과 연결된 가중치의 norm을 이용하여 학습하였다. 시뮬레이션을 통하여 기존 에러 역전파 신경망 학습과 그 성능을 비교한 결과 제안한 초기 가중치 표현이 학습속도와 수렴능력에서 우수함을 나타낸다.

• IEIE Transactions on Smart Processing and Computing
• /
• v.4 no.1
• /
• pp.35-43
• /
• 2015

Over the past couple of years, tremendous progress has been made in applying deep learning (DL) techniques to computer vision. Especially, deep convolutional neural networks (DCNNs) have achieved state-of-the-art performance on standard recognition datasets and tasks such as ImageNet Large-Scale Visual Recognition Challenge (ILSVRC). Among them, GoogLeNet network which is a radically redesigned DCNN based on the Hebbian principle and scale invariance set the new state of the art for classification and detection in the ILSVRC 2014. Since there exist various deep learning techniques, this review paper is focusing on techniques directly related to DCNNs, especially those needed to understand the architecture and techniques employed in GoogLeNet network.

• Journal of the Korean Institute of Telematics and Electronics C
• /
• v.34C no.6
• /
• pp.71-78
• /
• 1997

In this paper, a new neural network training algorithm will be devised for function approximator with good generalization characteristics and tested with the time series prediction problem using santaFe competition data sets. To enhance the generalization ability a constraint term of hidden neuraon activations is added to the conventional output error, which gives the curvature smoothing characteristics to multi-layer neural networks. A hybrid learning algorithm of the error-back propagation and Hebbian learning algorithm with weight decay constraint will be naturally developed by the steepest decent algorithm minimizing the proposed cost function without much increase of computational requriements.

• Journal of KIISE:Software and Applications
• /
• v.31 no.9
• /
• pp.1171-1179
• /
• 2004

We describe word boundary detection that extracts the boundary between speech and non-speech. The proposed method uses two features. One is the normalized root mean square of speech signal, which is insensitive to white noises and represents temporal information. The other is the normalized met-frequency band energy of voice signal, which is frequency information of the signal. Our method detects word boundaries using a recurrent fuzzy associative memory(RFAM) that extends FAM by adding recurrent nodes. Hebbian learning method is employed to establish the degree of association between an input and output. An error back-propagation algorithm is used for teaming the weights between the consequent layer and the recurrent layer. To confirm the effectiveness, we applied the suggested system to voice data obtained from KAIST.

• Journal of the Korean Data and Information Science Society
• /
• v.7 no.1
• /
• pp.113-118
• /
• 1996

Principal component analysis(PCA) is an essential technique for data compression and feature extraction, and has been widely used in statistical data analysis, communication theory, pattern recognition, and image processing. Oja(1992) found that a linear neuron with constrained Hebbian learning rule can extract the principal component by using stochastic gradient ascent method. In practice real data often contain some outliers. These outliers will significantly deteriorate the performances of the PCA algorithms. In order to make PCA robust, Xu & Yuille(1995) applied statistical physics to the problem of robust principal component analysis(RPCA). Devlin et.al(1981) obtained principal components by using techniques such as M-estimation. The propose of this paper is to investigate from the statistical point of view how Xu & Yuille's(1995) RPCA works under the same simulation condition as in Devlin et.al(1981).

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2001.12a
• /
• pp.181-184
• /
• 2001

인간 두뇌의 연상과 기억 작용의 모델링을 통한 구현의 일부분으로, 본 논문에서는 Hebb 의 학습방법과 non-cloning template를 사용하여 discrete-time cellular neural networks의 연상메모리 기능을 구현한다. 본 논문에서 사용된 학습방법은 각 셀의 인접한 셀과의 연결상태에 따라 하중값 메트릭스를 구현한다. 이러한 방법은 새로운 패턴의 추가 학습과 삭제가 쉽고, 또한 쉽게 구현 할 수 있는 장점이 있다. 이 방법으로 모의 실험에서는 교통표지판의 분류에 사용한다.