• Journal of KIISE:Software and Applications
• /
• v.26 no.12
• /
• pp.1383-1392
• /
• 1999

이미 학습된 다층퍼셉트론 신경망을 디지털 VLSI 기술을 사용하여 하드웨어로 구현할 경우 신경망의 가중치 및 뉴런 출력들을 양자화해야 하는 문제가 발생한다. 이러한 신경망 변수들의 양자화는 결과적으로 주어진 입력에 대한 신경망의 최종 출력에서의 왜곡을 초래한다. 본 논문에서는 먼저 이러한 양자화로 인한 신경망 출력에서의 왜곡을 통계적으로 분석하였다. 분석 결과에 의하면 입력패턴 각 성분의 제곱들의 합과 가중치의 크기들이 양자화 영향에 주로 기여하는 것으로 나타났다. 이러한 분석 결과를 이용하여 양자화를 위한 정밀도가 주어졌을 때, 양자화 영향이 최소화된 다층퍼셉트론 신경망을 설계하는 방법을 제시하였다. 그리고 제안된 방법에 의해 얻은 신경망과 오류역전파 학습방법에 의하여 얻은 신경망의 성능을 비교함으로써 제안된 방법의 효율성을 입증하였다. 실험결과는 낮은 양자화 정밀도에서도 제안된 방법이 더 좋은 성능을 보였다.Abstract When we implement a multilayer perceptron with the digital VLSI technology, we generally have to quantize the weights and the neuron outputs. These quantizations eventually cause distortion in the output of the network for a given input. In this paper first we made a statistical analysis about the effect caused by the quantization on the output of the network. The analysis revealed that the sum of the squared input components and the sizes of the weights are the major factors which contribute to the quantization effect. We present a design method for an MLP which minimizes the quantization effect when the precision of the quantization is given. In order to show the effectiveness of the proposed method, we developed a network by our method and compared it with the one developed by the regular backpropagation. We could confirm that the network developed by our method performs better even with a low precision of the quantization.

• Journal of the Korea Society of Computer and Information
• /
• v.15 no.10
• /
• pp.105-112
• /
• 2010

In this paper, the bio-signals, such as EEG, ECG were measured with a sensor and their characters were drawn out and analyzed. With results from the analysis, four emotion of rest, concentration, tension and depression were inferred. In order to assess one's emotion, the characteristic vectors were drawn out by applying various ways, including the frequency analysis of the bio-signals like the measured EEG and HRV. RBFN, a neural network of the complex structure of unsupervised and supervised learning, was applied to classify and infer the deducted information. Through experiments, the system suggested in this thesis showed better capability to classify and infer than other systems using a different neural network. As follow-up research tasks, the recognizance rate of the measured bio-signals should be improved. Also, the technology which can be applied to the wired or wireless sensor measuring the bio-signals more easily and to wearable computing should be developed.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.48 no.6
• /
• pp.43-50
• /
• 2011

Currently, conventional security methods including IC card or password type method are quickly switched into biometric security systems in various applications and the electrocardiogram (ECG) has been considered as one of novel biometrics way. However, conventional ECG based biometrics used lead II signal which conventionally used for formulaic signal to heart disease diagnosis and it is not suitable for biometrics since it is rather difficult to find consistent features for heart disease patents. To overcome this problem, we developed new biometrics system using ECG lead III signals. For wave extraction, signal peak points are extracted through AAV algorithm. For feature selection, extracted waves are categorized into one of four wave types and total twenty two features including number of vertices, wave shapes, amplitude information and interval information are extracted based on their wave types. Experimental results for thirty-six people showed 100% specificity, 95.59% sensitivity and 99.17% of overall identification accuracy.

• Journal of the Korea Society of Computer and Information
• /
• v.16 no.9
• /
• pp.27-33
• /
• 2011

In this paper, we propose an efficient on-line handwritten digit recognition base on Convex-Concave curves feature which is extracted by a chain code sequence using Smith-Waterman alignment algorithm. The time sequential signal from mouse movement on the writing pad is described as a sequence of consecutive points on the x-y plane. So, we can create data-set which are successive and time-sequential pixel position data by preprocessing. Data preprocessed is used for Convex-Concave curves feature extraction. This feature is scale-, translation-, and rotation-invariant. The extracted specific feature is fed to a Smith-Waterman alignment algorithm, which in turn classifies it as one of the nine digits. In comparison with backpropagation neural network, Smith-Waterman alignment has the more outstanding performance.

• Journal of Advanced Navigation Technology
• /
• v.13 no.6
• /
• pp.901-906
• /
• 2009

This paper proposes a call admission control(CAC) method for wireless networks, which is based on the upper bound of a possibility distribution of handoff calls dropping rates. The possibility distribution is estimated in a fuzzy inference and a learning algorithm in neural network. The learning algorithm is considered for tuning the membership functions(then parts)of fuzzy rules for the inference. The fuzzy inference method is based on a weighted average of fuzzy sets. The proposed method can avoid estimating excessively large handoff calls dropping rates, and makes possibile self-compensation in real time for the case where the estimated values are smaller than real values. So this method makes secure CAC, thereby guaranteeing the allowed CDR. From simulation studies we show that the estimation performance for the upper bound of call dropping rate is good, and then handoff call dropping rates in CAC are able to be sustained below user's desired value.

• The KIPS Transactions:PartB
• /
• v.13B no.6 s.109
• /
• pp.615-624
• /
• 2006

The objective of this paper is to implement parallel multi-layer ANN(Artificial Neural Network) simulator based on the mobile agent system which is executed in parallel in the virtual parallel distributed computing environment. The Multi-Layer Neural Network is classified by training session, training data layer, node, md weight in the parallelization-level. In this study, We have developed and evaluated the simulator with which it is feasible to parallel the ANN in the training session and training data parallelization because these have relatively few network traffic. In this results, we have verified that the performance of parallelization is high about 3.3 times in the training session and training data. The great significance of this paper is that the performance of ANN's execution on virtual parallel computer is similar to that of ANN's execution on existing super-computer. Therefore, we think that the virtual parallel computer can be considerably helpful in developing the neural network because it decreases the training time which needs extra-time.

• Journal of the Korea Society of Computer and Information
• /
• v.16 no.8
• /
• pp.137-146
• /
• 2011

In order to support increased consumer awareness regarding energy consumption, we present new ways of monitoring and predicting with energy in electric appliances. The proposed system is a design of a common electrical power outlet called smart plug that measures the amount of current passing through current sensor at 0.5 second. To acquire data for training and testing the proposed neural network, weather parameters used include average temperature of day, min and max temperature, humidity, and sunshine hour as input data, and power consumption as target data from smart plug. Using the experimental data for training, the neural network model based on Back-Propagation algorithm was developed. Multi layer perception network was used for nonlinear mapping between the input and the output data. It was observed that the proposed neural network model can predict the power consumption quite well with correlation coefficient was 0.9965, and prediction mean square error was 0.02033.

• Journal of KIISE:Software and Applications
• /
• v.34 no.7
• /
• pp.585-594
• /
• 2007

The back propagation algorithm took a long time to learn the input patterns and was difficult to train the additional or repeated learning patterns. So Aleksander proposed the binary neural network which could overcome the disadvantages of BP Network. But it had the limitation of repeated learning and was impossible to extract a generalized pattern. In this paper, we proposed a dynamic 3 dimensional Neuro System which was consisted of a learning network which was based on weightless neural network and a feedback module which could accumulate the characteristic. The proposed system was enable to train additional and repeated patterns. Also it could be produced a generalized pattern by putting a proper threshold into each learning-net's discriminator which was resulted from learning procedures. And then we reused the generalized pattern to elevate the recognition rate. In the last processing step to decide right category, we used maximum response detector. We experimented using the MNIST database of NIST and got 99.3% of right recognition rate for training data.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.15 no.2
• /
• pp.1059-1065
• /
• 2014

Wind turbine system can obtain the maximum wind energy using torque control under the rated wind speed, and wind turbine power is controlled as the rated power using pitch control over the rated wind speed. In this paper, we present a method for wind turbine pitch controller using neural networks. The purpose of the pitch control is to control generator speed and power in the above rated wind speed. To improve the neural network pitch controller, the difference between a rated and current speed of generator has been used for another input of neural networks as well as wind speed. Error back-propagation algorithm is used for training the neural network pitch controller and simulation and Matlab/Simulink is used for verifying that this system is controlled well.

• Journal of Korea Multimedia Society
• /
• v.11 no.7
• /
• pp.1035-1042
• /
• 2008

The support vector machine, as a binary classifier, is known to surpass the other classifiers only in binary classification problems through the various experiments. Even though its theory is based on the maximal margin classifier, the support vector machine approach cannot be easily extended to the multi-classification problems. In this paper, we review the extension techniques of the support vector machine toward the multi-classification and do the performance comparison. Depending on the data decomposition of the training data, the support vector machine is easily adapted for a multi-classification problem without modifying the intrinsic characteristics of the binary classifier. The performance is evaluated on a collection of the benchmark data sets and compared according to the selected teaming strategies, the training time, and the results of the neural network with the backpropagation teaming. The experiments suggest that the support vector machine is applicable and effective in the general multi-class classification problems when compared to the results of the neural network.