• Nuclear Engineering and Technology
• /
• v.39 no.4
• /
• pp.337-348
• /
• 2007

A fuzzy neural network model is presented to predict residual stress for dissimilar metal welding under various welding conditions. The fuzzy neural network model, which consists of a fuzzy inference system and a neuronal training system, is optimized by a hybrid learning method that combines a genetic algorithm to optimize the membership function parameters and a least squares method to solve the consequent parameters. The data of finite element analysis are divided into four data groups, which are split according to two end-section constraints and two prediction paths. Four fuzzy neural network models were therefore applied to the numerical data obtained from the finite element analysis for the two end-section constraints and the two prediction paths. The fuzzy neural network models were trained with the aid of a data set prepared for training (training data), optimized by means of an optimization data set and verified by means of a test data set that was different (independent) from the training data and the optimization data. The accuracy of fuzzy neural network models is known to be sufficiently accurate for use in an integrity evaluation by predicting the residual stress of dissimilar metal welding zones.

• The Transactions of the Korea Information Processing Society
• /
• v.4 no.11
• /
• pp.2769-2779
• /
• 1997

This Paper Proposes an efficient method for improving the training performance of the neural networks using a hybrid of conjugate gradient backpropagation algorithm and dynamic tunneling backpropagation algorithm The conjugate gradient backpropagation algorithm, which is the fast gradient algorithm, is applied for high speed optimization. The dynamic tunneling backpropagation algorithm, which is the deterministic method with tunneling phenomenon, is applied for global optimization. Conversing to the local minima by using the conjugate gradient backpropagation algorithm, the new initial point for escaping the local minima is estimated by dynamic tunneling backpropagation algorithm. The proposed method has been applied to the parity check and the pattern classification. The simulation results show that the performance of proposed method is superior to those of gradient descent backpropagtion algorithm and a hybrid of gradient descent and dynamic tunneling backpropagation algorithm, and the new algorithm converges more often to the global minima than gradient descent backpropagation algorithm.

• Journal of the Korean Data and Information Science Society
• /
• v.23 no.5
• /
• pp.1027-1035
• /
• 2012

This paper proposes a new Levenberg-Marquardt algorithm that is accelerated by adjusting a Jacobian matrix and a quasi-Hessian matrix. The proposed method partitions the Jacobian matrix into block matrices and employs the inverse of a partitioned matrix to find the inverse of the quasi-Hessian matrix. Our method can avoid expensive operations and save memory in calculating the inverse of the quasi-Hessian matrix. It can shorten the training time for fast convergence. In our results tested in a large application, we were able to save about 20% of the training time than other algorithms.

• 제어로봇시스템학회:학술대회논문집
• /
• 2004.08a
• /
• pp.1413-1415
• /
• 2004

This paper proposes a SVM(Support Vector Machine) training algorithm to control a service robot with voice command. The service robot with a stereo vision system and dual manipulators of four degrees of freedom implements a User-Dependent Voice Control System. The training of SVM algorithm that is one of the statistical learning theories leads to a QP(quadratic programming) problem. In this paper, we present an efficient SVM speech recognition scheme especially based on less learning data comparing with conventional approaches. SVM discriminator decides rejection or acceptance of user's extracted voice features by the MFCC(Mel Frequency Cepstrum Coefficient). Among several SVM kernels, the exponential RBF function gives the best classification and the accurate user recognition. The numerical simulation and the experiment verified the usefulness of the proposed algorithm.

• 제어로봇시스템학회:학술대회논문집
• /
• 2004.08a
• /
• pp.185-189
• /
• 2004

For face recognition system, a face detector which can find exact face region from complex image is needed. Many face detection algorithms have been developed under the assumption that background of the source image is quite simple . this means that face region occupy more than a quarter of the area of the source image or the background is one-colored. Color-based face detection is fast but can't be applicable to the images of which the background color is similar to face color. And the algorithm using neural network needs so many non-face data for training and doesn't guarantee general performance. In this paper, A multi-scale, multi-face detection algorithm using PCA is suggested. This algorithm can find most multi-scaled faces contained in static images with small number of training data in reasonable time.

• Proceedings of the KSPS conference
• /
• 2002.11a
• /
• pp.183-186
• /
• 2002

The speech detection is one of the important problems in real-time speech recognition. The accurate detection of speech boundaries is crucial to the performance of speech recognizer. In this paper, we propose a speech detector based on Mel-band selection through training. In order to show the excellence of the proposed algorithm, we compare it with a conventional one, so called, EPD-VAA (EndPoint Detector based on Voice Activity Detection). The proposed speech detector is trained in order to better extract keyword speech than other speech. EPD-VAA usually works well in high SNR but it doesn't work well any more in low SNR. But the proposed algorithm pre-selects useful bands through keyword training and decides the speech boundary according to the energy level of the sub-bands that is previously selected. The experimental result shows that the proposed algorithm outperforms the EPD-VAA.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1992.06a
• /
• pp.61-65
• /
• 1992

This paper presents a study on Neural Networks for Phoneme Recognition and performs phoneme recognition using TDNN(Time Delay Neural Network). Also, this paper proposes new training algorithm for speech recognition using neural nets that proper to large scale TDNN. Because phoneme recognition is indispensable for continuous speech recognition, this paper uses TDNN to get accurate recognition result of phoneme. And this paper proposes new training algorithm that can converge TDNN to optimal state regardless of the number of phoneme to be recognized. The result of recognition on three phoneme classes shows recognition rate of 9.1%. And this paper proves that proposed algorithm is a efficient method for high performance and reducing convergence time.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2009.01a
• /
• pp.254-257
• /
• 2009

In this paper, we propose a learning-based super-resolution algorithm. In the proposed algorithm, a multi-resolution wavelet approach is adopted to perform the synthesis of local high-frequency features. To obtain a high-resolution image, wavelet coefficients of two dominant LH- and HL-bands are estimated based on wavelet frames. In order to prepare more efficient training sets, the proposed algorithm utilizes the LH-band and transposed HL-band. The training sets are then used for the estimation of wavelet coefficients for both LH- and HL-bands. Using the estimated high frequency bands, a high resolution image is reconstructed via the wavelet transform. Experimental results demonstrate that the proposed scheme can synthesize high-quality images.

• Journal of the Korean Society of Manufacturing Technology Engineers
• /
• v.18 no.4
• /
• pp.430-435
• /
• 2009

The cooling stage greatly affects the product quality in the injection molding process. The cooling system that minimizes temperature variance in the product surface will improve the quality and the productivity of products. The cooling circuit optimization problem that was once solved by a response surface method with 4 design variables. It took too much time for the optimization as an industrial design tool. It is desirable to reduce the optimization time. Therefore, we tried the back-propagation algorithm of artificial neural network(BPN) to find an optimum solution in the cooling circuit design in this research. We tried various ways to select training points for the BPN. The same optimum solution was obtained by applying the BPN with reduced number of training points by the fractional factorial design.

• Proceedings of the Korean Society of Surveying, Geodesy, Photogrammetry, and Cartography Conference
• /
• 2004.11a
• /
• pp.251-256
• /
• 2004

The image classification is one of the most important studies in the remote sensing. In general, the MLC(Maximum Likelihood Classification) classification that in consideration of distribution of training information is the most effective way but it produces a bad result when we apply it to actual hyperspectral image with the same classification technique. The purpose of this research is to reveal that which one is the most effective and suitable way of the classification algorithms iii the hyperspectral image classification. To confirm this matter, we apply the MLC classification algorithm which has distribution information and SAM(Spectral Angle Mapper), SFF(Spectral Feature Fitting) algorithm which use average information of the training class to both multispectral image and hyperspectral image. I conclude this result through quantitative and visual analysis using confusion matrix could confirm that SAM and SFF algorithm using of spectral pattern in vector domain is more effective way in the hyperspectral image classification than MLC which considered distribution.