• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1993.06a
• /
• pp.857-860
• /
• 1993

In this paper, we apply the self generating neuro fuzzy model (SGNFM) to the dimension analysis of the chaotic time series. Firstly, we formulate a nonlinear time series identification problem with nonlinear autoregressive (NARMAX) model. Secondly, we propose an identification algorithm using SGNFM. We apply this method to the estimation of embedding dimension for chaotic time series, since the embedding dimension plays an essential role for the identification and the prediction of chaotic time series. In this estimation method, identification problems with gradually increasing embedding dimension are solved, and the identified result is used for computing correlation coefficients between the predicted time series and the observed one. We apply this method to the dimension estimation of a chaotic pulsation in a finger's capillary vessels.

• Elastomers and Composites
• /
• v.54 no.2
• /
• pp.156-160
• /
• 2019

The changes in the dispersion of carbon black in liquid polyisoprene under shear flow with time have been investigated by time-resolved ultra small-angle X-ray scattering (USAXS) method. The analyses of USAXS profile immediately after the start of shear flow clarified that the aggregates of carbon black with a mean radius of gyration of 14 nm and surface fractal dimension of 2.5 form the fractal network structure with mass-fractal dimension of 2.9. After the application of the shear flow, the scattering intensity increases with time at the observed whole entire q region, and then the a shoulder appears at $q=0.005nm^{-1}$, indicating that the agglomerate is broken and becomes smaller by shear flow. The analysis by the Unified Guinier/Power-law approach yielded several characteristic parameters, such as the sizes of aggregate and agglomerate, mass-fractal dimension of agglomerate, and surface fractal dimension of the primary particle. While the mean radius of gyration of the agglomerate decreases with time, the mean radius of gyration of the aggregate, mass fractal dimension, and surface fractal dimension don't change with time, indicating that the aggregates peel off the surface of the agglomerate.

• Speech Sciences
• /
• v.3
• /
• pp.148-155
• /
• 1998

As an application of dimensional analysis in the theory of chaos and fractals, we studied and estimated the information dimension for various phonemes. By constructing phase-space vectors from the time-series speech signals, we calculated the natural measure and the Shannon's information from the trajectories. The information dimension was finally obtained as the slope of the plot of the information versus space division order. The information dimension showed that it is so sensitive to the waveform and time delay. By averaging over frames for various phonemes, we found the information dimension ranges from 1.2 to 1.4.

• Communications for Statistical Applications and Methods
• /
• v.31 no.2
• /
• pp.213-234
• /
• 2024

In recent decades, increasing research attention has been directed toward predicting the price of stocks in financial markets using deep learning methods. For instance, recurrent neural network (RNN) is known to be competitive for datasets with time-series data. Long short term memory (LSTM) further improves RNN by providing an alternative approach to the gradient loss problem. LSTM has its own advantage in predictive accuracy by retaining memory for a longer time. In this paper, we combine both supervised and unsupervised dimension reduction methods with LSTM to enhance the forecasting performance and refer to this as a dimension reduction based LSTM (DR-LSTM) approach. For a supervised dimension reduction method, we use methods such as sliced inverse regression (SIR), sparse SIR, and kernel SIR. Furthermore, principal component analysis (PCA), sparse PCA, and kernel PCA are used as unsupervised dimension reduction methods. Using datasets of real stock market index (S&P 500, STOXX Europe 600, and KOSPI), we present a comparative study on predictive accuracy between six DR-LSTM methods and time series modeling.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.15 no.7 s.100
• /
• pp.781-787
• /
• 2005

The correlation dimension can provide some intrinsic Information of an underlying dynamic system by reconstructing measured nonlinear time series. The vibration signals measured from a rotor with different clearance sizes between shaft and bushing were analyzed using the correlation dimension. The results showed that the correlation dimension can identify the size of the clearance of a rotor and the lubricating condition, which can not be analyzed by frequency spectrum or wavelet. The magnitude of the correlation dimension became smaller as the clearance larger and as the lubrication condition better.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2004.11a
• /
• pp.134-139
• /
• 2004

The correlation dimension of a nonlinear method for the diagnosis on the clearance of rotating machinery is introduced in this paper. The correlation dimension can provide some intrinsic information of an underlying dynamic system by reconstructing measured scalar time series. Vibration signals measured from a rotor with different operating conditions are analyzed using the correlation dimension. The results show that the correlation dimension method can identify the magnitude of the clearance of a rotor and the lubricating condition.

• Speech Sciences
• /
• v.4 no.2
• /
• pp.25-39
• /
• 1998

This paper deals with fuzzy correlation dimension for an appropriate speech recognition. The proposed fuzzy correlation dimension has absorbed time variation value of strange attractor as utilizing fuzzy membership function at calculation of integral correlation when the results of proposed dimension are applied to speech recognition fuzzed correlation dimension is superior to speech recognition, and correlation dimension is superior to speaker discrimination.

• Korean Journal of Veterinary Research
• /
• v.37 no.3
• /
• pp.669-685
• /
• 1997

The present study was performed to evaluate the effects of xylazine and tiletamine + zolazepam on echocardiograms before and after experimental myocardial infarctions in clinically normal dogs taken preliminary examinations related to cardiac function. The results are as follows. With xylazine administration, left ventricle end-diastolic dimension, left ventricle end-systolic dimension, left atrium/aorta, ejection time and velocity of circumferential fiber shortening increased and mitral valve CD slope, % delta D decreased(p<0.01). In tiletamine+zolazepam administered group, interventricular septum amplitude(p<0.01), mitral valve DE slope(p<0.05) and ejection time(p<0.01) decreased and left atrium/aorta, ejection time also decreased compared with xylazine group(p<0.01). In 48 hours after experimental myocardial infarction group, anterior aortic wall amplitude decreased compared with control, xylazine, tiletamine + zolazepam group, respectively(p<0.01). Posterior aortic wall amplitude decreased compared with control(p<0.01). Left ventricle end systolic dimension increased compared with control and tiletamine + zolazepam group, respectively(p<0.01). Left ventricular posterior wall end systolic dimension decreased compared with control(p<0.01). Left ventricular posterior wall amplitude decreased compared with control and tiletamine+zolazepam group(p<0.01). Left atrium/aorta decreased compared with xylazine group(p<0.01). % thickening left ventricular posterior wall decreased compared with control(p<0.05). % delta D decreased compared with control and tiletamine+zolazepam group(p<0.01). Ejection time decreased compared with xylazine(p<0.01). Velocity of circumferential fiber shortening increased compared with control and tiletamine + zolazepam group(p<0.01). With xylazine administration 48 hours after experimental myocardial infarction, anterior aortic wall amplitude, posterior aortic wall amplitude decreased compared with control(p<0.01). Left ventricle end-diastolic dimension increased compared with control(p<0.01). Left ventricle end-systolic dimension increased compared with control and tiletamine + zolazepam group, respectively(p<0.01). Left ventricular posterior wall end-systolic dimension and left ventricular posterior wall end-diastolic dimension decreased compared with control(p<0.01). Left atrium/aorta decreased compared with xylazine group(p<0.01). % thickening left ventricular posterior. wall(p<0.05) and % delta D(p<0.01) decreased compared with control. Velocity of circumferential fiber shortening increased compared with tiletamine + zolazepam group(p<0.01). With tiletamine + zolazepam administration 48 hours after experimental myocardial infarction, anterior aortic wall amplitude decreased compared with control, xylazine and tiletamine+zolazepam group, respectively(p<0.01). Posterior aortic wall amplitude decreased compared with control(p<0.01). Left ventricle end-systolic dimension increased compared with control and tiletamine+zolazepam group(p<0.01). Left ventricular posterior wall end-systolic dimension, left ventricular posterior wall end-diastolic dimension and interventricular septum amplitude decreased compared with control(p<0.01). Left atrium/aorta decreased compared with xylazine group(p<0.01). % delta D decreased compared with control and tiletamine + zolazepam group(p<0.01). Ejection time decreased compared with xylazine group and velocity of circumferential fiber shortening increased compared withtiletamine+zolazepam group(p<0.01). Conclusively, echocardiography was proved to be a useful, diagnostic, non-invasive and simple method for establishing the diagnosis of myocardial infarction and evaluating the effects of drug on cardiac function before and after myocardial infarction.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.15 no.2
• /
• pp.121-125
• /
• 2015

In this paper, we propose a method that extracts features from character patterns using the fractal dimension of chaos theory. The input character pattern image is converted into time-series data. Then, using the modified Henon system suggested in this paper, it determines the last features of the character pattern image after calculating the box-counting dimension, natural measure, information bit, and information (fractal) dimension. Finally, character pattern recognition is performed by statistically finding each information bit that shows the minimum difference compared with a normalized character pattern database.

• The Korean Journal of Applied Statistics
• /
• v.23 no.5
• /
• pp.813-822
• /
• 2010

We investigate a possible achievement in dimension reduction of time series via partially quantified principal component. Partial quantification technique allows us in modeling to accommodate artificial variable(s) of practical importance which is defined subjectively by the data analyst. Suggested procedures are described and in turn illustrated in detail by analyzing monthly unemployment rates in Korea.