• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2007.05a
• /
• pp.588-593
• /
• 2007

Fiber reinforced polymer(FRP) composite decks are new to bridge applications and hence not much literature exists on their structural mechanical behavior. As there are many differences between numerical displacements through static analysis of the primary model and experimental displacements through static load tests, system identification (SI)techniques such as Neural Networks (NN) and support vector machines (SVM) utilized in the optimization of the FE model. During the process of identification, displacements were used as input while stiffness as outputs. Through the comparison of numerical displacements after SI and experimental displacements, it can note that NN and SVM would be effective SI methods in modeling an FRP deck. Moreover, two methods such as response surface method and iteration were proposed to optimize the estimated stiffness. Finally, the results were compared through the mean square error (MSE) of the differences between numerical displacements and experimental displacements at 6 points.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 2007.04a
• /
• pp.565-570
• /
• 2007

Fiber reinforced polymer(FRP) composite decks are new to bridge applications and hence not much literature exists on their structural mechanical behavior. As there are many differences between numerical displacements through static analysis of the primary model and experimental displacements through static load tests, system identification (SI)techniques such as Neural Networks (NN) and support vector machines (SVM) utilized in the optimization of the FE model. During the process of identification, displacements were used as input while stiffness as outputs. Through the comparison of numerical displacements after SI and experimental displacements, it can note that NN and SVM would be effective SI methods in modeling an FRP deck. Moreover, two methods such as response surface method and iteration were proposed to optimize the estimated stiffness. Finally, the results were compared through the mean square error (MSE) of the differences between numerical displacements and experimental displacements at 6 points.

• The Journal of Korean Institute of Next Generation Computing
• /
• v.15 no.3
• /
• pp.16-24
• /
• 2019

Recently, research on emotion recognition based on EEG has attracted great interest from human-robot interaction field. In this paper, we propose a method of labeling using image-based EEG topography instead of evaluating emotions through self-assessment and annotation labeling methods used in MAHNOB HCI. The proposed method evaluates the emotion by machine learning model that learned EEG signal transformed into topographical image. In the experiments using MAHNOB-HCI database, we compared the performance of training EEG topography labeling models of SVM and kNN. The accuracy of the proposed method was 54.2% in SVM and 57.7% in kNN.

• Journal of Korea Multimedia Society
• /
• v.9 no.11
• /
• pp.1465-1473
• /
• 2006

In this paper, we present an approach for facial expression recognition using Active Shape Models(ASM) and a state-based model in image sequences. Given an image frame, we use ASM to obtain the shape parameter vector of the model while we locate facial feature points. Then, we can obtain the shape parameter vector set for all the frames of an image sequence. This vector set is converted into a state vector which is one of the three states by the state-based model. In the classification step, we use the k-NN with the proposed similarity measure that is motivated on the observation that the variation-regions of an expression sequence are different from those of other expression sequences. In the experiment with the public database KCFD, we demonstrate that the proposed measure slightly outperforms the binary measure in which the recognition performance of the k-NN with the proposed measure and the existing binary measure show 89.1% and 86.2% respectively when k is 1.

• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.34 no.6
• /
• pp.315-324
• /
• 2022

The experimental data obtained in a wave flume were analyzed using machine learning techniques to establish a model that predicts the input wave height of the wavemaker based on the waves that have experienced wave shoaling and to verify the performance of the established model. For this purpose, artificial neural network (NN), the most representative machine learning technique, and Gaussian process regression (GPR), one of the non-parametric regression analysis methods, were applied respectively. Then, the predictive performance of the two models was compared. The analysis was performed independently for the case of using all the data at once and for the case by classifying the data with a criterion related to the occurrence of wave breaking. When the data were not classified, the error between the input wave height at the wavemaker and the measured value was relatively large for both the NN and GPR models. On the other hand, if the data were divided into non-breaking and breaking conditions, the accuracy of predicting the input wave height was greatly improved. Among the two models, the overall performance of the GPR model was better than that of the NN model.

• Journal of Power Electronics
• /
• v.6 no.1
• /
• pp.38-44
• /
• 2006

In this paper, a speed sensorless control for an ultrasonic motor (USM) using a neural network (NN) is presented. In the proposed method, rotor speed is estimated by a three-layer NN which adapts nonlinearities associated with load torque and motor temperature into control. The intrinsic properties of a USM, such as high torque for low speeds, high static torque, compact size, etc., offer great advantages for industrial applications. However, the speed property of a USM has strong nonlinear properties associated with motor temperature and load torque, which make accurate speed control difficult. These properties are considered in designing a control method through the application of mathematical models. In these strategies, a detailed speed model of the USM is required which makes actual applications impractical. In the proposed method, a three-layer NN estimates the speed of the USM from the drive frequency, the root mean square value of input voltage and the surface temperature of the USM, where no mechanical speed sensor is needed. The NN speed based estimator enables inclusion of variations in driving conditions due to input signals of the NN involved during the driving state of the USM. The disuse of sensors offers many advantages on both the cost and maintenance front. Moreover, the model free sensorless control method offers practical controller construction within a small number of parameters. To validate the proposed speed sensorless control method for a USM, experiments have been executed under several conditions.

• Atmosphere
• /
• v.30 no.4
• /
• pp.405-420
• /
• 2020

In this study, we investigate the predictability of intraseasonal monthly temperature and precipitation variations using hindcast datasets from eight global circulation models participating in the operational multi-model ensemble (MME) seasonal prediction system of the Asia-Pacific Economic Cooperation Climate Center for the 1983~2010 period. These intraseasonal monthly variations are defined by categorical deterministic analysis. The monthly temperature and precipitation are categorized into above normal (AN), near normal (NN), and below normal (BN) based on the σ-value ± 0.43 after standardization. The nine patterns of intraseasonal monthly variation are defined by considering the changing pattern of the monthly categories for the three consecutive months. A deterministic and a probabilistic analysis are used to define intraseasonal monthly variation for the multi-model consisting of numerous ensemble members. The results show that a pattern (pattern 7), which has the same monthly categories in three consecutive months, is the most frequently occurring pattern in observation regardless of the seasons and variables. Meanwhile, the patterns (e.g., patterns 8 and 9) that have consistently increasing or decreasing trends in three consecutive months, such as BN-NN-AN or AN-NN-BN, occur rarely in observation. The MME and eight individual models generally capture pattern 7 well but rarely capture patterns 8 and 9.

• Geomechanics and Engineering
• /
• v.33 no.2
• /
• pp.183-194
• /
• 2023

This study presents the prediction of the California bearing ratio (CBR) of coarse- and fine-grained soils using artificial intelligence technology. The group method of data handling (GMDH) algorithm, an artificial neural network-based model, was used in the prediction of the CBR values. In the design of the prediction models, various combinations of independent input variables for both coarse- and fine-grained soils have been used. The results obtained from the designed GMDH-type neural networks (GMDH-type NN) were compared with other regression models, such as linear, support vector, and multilayer perception regression methods. The performance of models was evaluated with a regression coefficient (R²), root-mean-square error (RMSE), and mean absolute error (MAE). The results showed that GMDH-type NN algorithm had higher performance than other regression methods in the prediction of CBR value for coarse- and fine-grained soils. The GMDH model had an R² of 0.938, RMSE of 1.87, and MAE of 1.48 for the input variables {G, S, and MDD} in coarse-grained soils. For fine-grained soils, it had an R² of 0.829, RMSE of 3.02, and MAE of 2.40, when using the input variables {LL, PI, MDD, and OMC}. The performance evaluations revealed that the GMDH-type NN models were effective in predicting CBR values of both coarse- and fine-grained soils.

• Journal of the Korean Data and Information Science Society
• /
• v.20 no.6
• /
• pp.1085-1092
• /
• 2009

The ozone data is one of the important environmental data for measurement of the atmospheric condition of the country. In this article, the Autoregressive Error (ARE) model and Neural Networks (NN) model have been considered for analyzing the ozone data at the northern part of the Gyeonggi-Do, Paju monitoring site in Korea. In the both ARE model and NN model, seven meteorological variables and four pollution variables are used as the explanatory variables for the ozone data set. The seven meteorological variables are daily maximum temperature, wind speed, relative humidity, rainfall, dew point temperature, steam pressure, and amount of cloud. The four air pollution explanatory variables are Sulfur dioxide ($SO_2$), Nitrogen dioxide ($NO_2$), Cobalt (CO), and Promethium 10 (PM10). The result showed that the NN model is generally better suited for describing the ozone concentration than the ARE model. However, the ARE model will be expected also good when we add the explanatory variables in the model.

• Journal of Intelligence and Information Systems
• /
• v.25 no.3
• /
• pp.239-251
• /
• 2019

Statistical methods such as moving averages, Kalman filtering, exponential smoothing, regression analysis, and ARIMA (autoregressive integrated moving average) have been used for stock market predictions. However, these statistical methods have not produced superior performances. In recent years, machine learning techniques have been widely used in stock market predictions, including artificial neural network, SVM, and genetic algorithm. In particular, a case-based reasoning method, known as k-nearest neighbor is also widely used for stock price prediction. Case based reasoning retrieves several similar cases from previous cases when a new problem occurs, and combines the class labels of similar cases to create a classification for the new problem. However, case based reasoning has some problems. First, case based reasoning has a tendency to search for a fixed number of neighbors in the observation space and always selects the same number of neighbors rather than the best similar neighbors for the target case. So, case based reasoning may have to take into account more cases even when there are fewer cases applicable depending on the subject. Second, case based reasoning may select neighbors that are far away from the target case. Thus, case based reasoning does not guarantee an optimal pseudo-neighborhood for various target cases, and the predictability can be degraded due to a deviation from the desired similar neighbor. This paper examines how the size of learning data affects stock price predictability through k-nearest neighbor and compares the predictability of k-nearest neighbor with the random walk model according to the size of the learning data and the number of neighbors. In this study, Samsung electronics stock prices were predicted by dividing the learning dataset into two types. For the prediction of next day's closing price, we used four variables: opening value, daily high, daily low, and daily close. In the first experiment, data from January 1, 2000 to December 31, 2017 were used for the learning process. In the second experiment, data from January 1, 2015 to December 31, 2017 were used for the learning process. The test data is from January 1, 2018 to August 31, 2018 for both experiments. We compared the performance of k-NN with the random walk model using the two learning dataset. The mean absolute percentage error (MAPE) was 1.3497 for the random walk model and 1.3570 for the k-NN for the first experiment when the learning data was small. However, the mean absolute percentage error (MAPE) for the random walk model was 1.3497 and the k-NN was 1.2928 for the second experiment when the learning data was large. These results show that the prediction power when more learning data are used is higher than when less learning data are used. Also, this paper shows that k-NN generally produces a better predictive power than random walk model for larger learning datasets and does not when the learning dataset is relatively small. Future studies need to consider macroeconomic variables related to stock price forecasting including opening price, low price, high price, and closing price. Also, to produce better results, it is recommended that the k-nearest neighbor needs to find nearest neighbors using the second step filtering method considering fundamental economic variables as well as a sufficient amount of learning data.