• The Journal of the Acoustical Society of Korea
• /
• v.20 no.6
• /
• pp.51-56
• /
• 2001

Recent researches have been focusing on jointly using lip motions and speech for reliable speech recognitions in noisy environments. To this end, this paper proposes the method of combining the visual speech recognizer and the conventional speech recognizer with each output properly weighted. In particular, we propose the method of autonomously determining the weights, depending on the amounts of noise in the speech. The correlations between adjacent speech samples and the residual errors of the LPC analysis are used for this determination. Simulation results show that the speech recognizer combined in this way provides the recognition performance of 83 ％ even in severely noisy environments.

• Journal of Powder Materials
• /
• v.26 no.5
• /
• pp.369-374
• /
• 2019

The properties of powder metallurgy products are related to their densities. In the present work, we demonstrate a method to apply artificial neural networks (ANNs) trained on experimental data to predict the bulk density of barium titanates. The density is modeled as a function of pressure, press rate, heating rate, sintering temperature, and soaking time using the ANN method. The model predictions with the training and testing data result in a high coefficient of correlation (R2 = 0.95 and Pearson's r = 0.97) and low average error. Moreover, a graphical user interface for the model is developed on the basis of the transformed weights of the optimally trained model. It facilitates the prediction of an infinite combination of process parameters with reasonable accuracy. Sensitivity analysis performed on the ANN model aids the identification of the impact of process parameters on the density of barium titanates.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2023.05a
• /
• pp.180-180
• /
• 2023

To improve global risk management, understanding the characteristics and distribution of precipitation is crucial. However, obtaining spatially and temporally resolved climatic data remains challenging due to sparse gauge observations and limited data availability, despite the use of satellite and reanalysis products. To address this challenge, merging available precipitation products has been introduced to generate spatially and temporally reliable data by taking advantage of the strength of the individual products. However, most of the existing studies utilize all the available products without considering the varying performances of each dataset in different regions. Comprehensively considering the relative contributions of each parent dataset is necessary since their contributions may vary significantly and utilizing all the available datasets for data merging may lead to significant data redundancy issues. Hence, for this study, we introduce a site-specific precipitation merging method that utilizes the Quadruple Collocation (QC) approach, which acknowledges the existence of error-cross correlation between the parent datasets, to create a high-resolution global daily precipitation data from 2001-2020. The performance of multiple gridded precipitation products are first evaluated per region to determine the best combination of quadruplets to be utilized in estimating the error variances through the QC approach and computation of merging weights. The merged precipitation is then computed by adding the precipitation from each dataset in the quadruplet multiplied by each respective merging weight. Our results show that our approach holds promise for generating reliable global precipitation data for data-scarce regions lacking spatially and temporally resolved precipitation data.

• Journal of Animal Science and Technology
• /
• v.46 no.5
• /
• pp.707-718
• /
• 2004

This study was conducted to evaluate the adequacy of an alternative a.m. - p.m. testing scheme for milk yield in comparison with the official test method based on weighing two milkings within 24 h. A total of 8,309 p.m. milking weights and 6,767 am. milking weights from 72 Holstein cows raised at N.L.R.I. were collected between October 2000 and November 2001. Seven statistical models were fitted to the data to derive formulas for estimating daily milk yields from morning or evening yields. In general, use of evening milkings less accurately estimated than did use morning rnilkings. Although the models do not differ much in the correlations between estimated and true daily milk yields, systematic under- and overestimation of daily milk yields were observed in all models with the exception of model 7, which accounted for heterogeneous variances by parity class, milking interval class, and lactation stage by fitting separate regression formulas within each combination of three factors.

• Journal of the Korean Society of Hazard Mitigation
• /
• v.9 no.3
• /
• pp.117-126
• /
• 2009

Various dimensions of watershed structural/non-structural planning can be applied in comprehensive flood mitigation plan in a river basin. Especially structural counterplans have very broad and diverse nature as flood control facilities. It is not easy to find the optimum alternative to maximize the ability of a basin to reduce flood risk using a combination of structural counterplans. In addition, there is no standard for evaluating the performance of structural counterplans and for selecting optimal combination of them. This study focused on how to select the best alternative of a comprehensive watershed structural plan from various flood defense alternative candidates. By introducing an analytic hierarchy process, we would like to show how we decide the best alternative using standard worksheets developed in this study for economics and policy evaluation, and Expert Choice 11.5, which calculates weights for evaluation items. Based on the results from this study, we would like to suggest the best practice of a standardized watershed plan for flood protection.

• Journal of Ocean Engineering and Technology
• /
• v.36 no.5
• /
• pp.313-325
• /
• 2022

Recently around the world, coastal erosion is paying attention as a social issue. Various constructions using low-crested and submerged structures are being performed to deal with the problems. In addition, a prediction study was researched using machine learning techniques to determine the wave attenuation characteristics of low crested structure to develop prediction matrix for wave attenuation coefficient prediction matrix consisting of weights and biases for ease access of engineers. In this study, a deep neural network model was constructed to predict the wave height transmission rate of low crested structures using Tensor flow, an open source platform. The neural network model shows a reliable prediction performance and is expected to be applied to a wide range of practical application in the field of coastal engineering. As a result of predicting the wave height transmission coefficient of the low crested structure depends on various input variable combinations, the combination of 5 condition showed relatively high accuracy with a small number of input variables defined as 0.961. In terms of the time cost of the model, it is considered that the method using the combination 5 conditions can be a good alternative. As a result of predicting the wave transmission rate of the trained deep neural network model, MSE was 1.3×10^-3, I was 0.995, SI was 0.078, and I was 0.979, which have very good prediction accuracy. It is judged that the proposed model can be used as a design tool by engineers and scientists to predict the wave transmission coefficient behind the low crested structure.

• Journal of the Korea Society of Computer and Information
• /
• v.11 no.3
• /
• pp.193-201
• /
• 2006

This paper presents a method that allows for detection of all rapid and gradual scene changes. The method features a combination of the current color histogram and the local $X^{2}-test$. For the purpose of this paper, the $X^{2}-test$ scheme outperforming existing histogram-based algorithms was transformed, and a local $X^{2}-test$ in which weights were applied in accordance with the degree of brightness was used to increase detection efficiency in the segmentation of color values. This Method allows for analysis and segmentation of complex time-varying images in the most general and standardized manner possible Experiments were performed to compare the proposed local $X^{2}-test$ method with the current $X^{2}-test$ method.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.9 no.9
• /
• pp.3293-3311
• /
• 2015

Cloud services are required to be composed as a single service to fulfill the workflow applications. Service composition in Cloud raises new challenges caused by the diversity of users with different QoS requirements and vague preferences, as well as the development of cloud computing having geographically distributed characteristics. So the selection of the best service composition is a complex problem and it faces trade-off among various QoS criteria. In this paper, we propose a Cloud service composition approach based on evolutionary algorithms, i.e., NSGA-II and MOPSO. We utilize the combination of multi-objective evolutionary approaches and Decision-Making method (AHP) to solve Cloud service composition optimization problem. The weights generated from AHP are applied to the Crowding Distance calculations of the above two evolutionary algorithms. Our algorithm beats single-objective algorithms on the optimization ability. And compared with general multi-objective algorithms, it is able to precisely capture the users' preferences. The results of the simulation also show that our approach can achieve a better scalability.

• Steel and Composite Structures
• /
• v.39 no.1
• /
• pp.81-93
• /
• 2021

Splitting tensile strength (STS) is an important mechanical parameter of concrete. This study offers novel methodologies for the early prediction of this parameter. Artificial neural network (ANN), which is a leading predictive method, is synthesized with two metaheuristic algorithms, namely atom search optimization (ASO) and equilibrium optimizer (EO) to achieve an optimal tuning of the weights and biases. The models are applied to data collected from the published literature. The sensitivity of the ASO and EO to the population size is first investigated, and then, proper configurations of the ASO-NN and EO-NN are compared to the conventional ANN. Evaluating the prediction results revealed the excellent efficiency of EO in optimizing the ANN. Accuracy improvements attained by this algorithm were 13.26 and 11.41% in terms of root mean square error and mean absolute error, respectively. Moreover, it raised the correlation from 0.89958 to 0.92722. This is while the results of the conventional ANN were slightly better than ASO-NN. The EO was also a faster optimizer than ASO. Based on these findings, the combination of the ANN and EO can be an efficient non-destructive tool for predicting the STS.

• Asia pacific journal of information systems
• /
• v.18 no.3
• /
• pp.1-20
• /
• 2008

Based on the multi-attribute utility theory (MAUT), we present a survey method to measure consumer preferences. The multi-attribute utility theory has been used to make decisions in OR/MS field; however, we show that the method can be effectively used to estimate the demand for new services by measuring individual level utility function. Because conjoint method has been widely used to measure consumer preferences for new products and services, we compare the pros and cons of two consumer preference survey methods. Further, we illustrate how swing weighing method can be effectively used to elicit customer preferences especially for new telecommunications services, Multi-attribute utility theory is a compositional approach for modeling customer preference, in which researchers calculate overall service utility by summing up the evaluation results for each attribute. On the contrary, conjoint method is a decompositional approach, which requires holistic evaluations for profiles. Partworth for each attribute is derived or estimated based on the evaluation, and finally consumer preferences for each profile are calculated. However, if the profiles are quite new and unfamiliar to the survey respondents, they will find it very difficult to accurately evaluate the profiles. We believe that the multi-attribute utility theory-based survey method is more appropriate than the conjoint method, because respondents only need to assess attribute level preferences and not holistic assessment. We chose swing weighting method among many weight assessment methods in multi-attribute utility theory, because it is designed to perform in a simple and fast manner. As illustrated in Clemen and Reilly (2001), to assess swing weights, the first step is to create the worst possible outcome as a benchmark by setting the worst level on each of the attributes. Then, each of the succeeding rows "swings" one of the attributes from worst to best. Upon constructing the swing table, respondents rank order the outcomes (rows). The next step is to rate the outcomes in which the rating for the benchmark is set to be 0 and the rating for the best outcome to be 100, and the ratings for other outcomes are determined in the ranges between 0 and 100. In calculating weight for each attribute, ratings are normalized by the total sum of all ratings. To demonstrate the applicability of the approach, we elicited and analyzed individual-level customer preference for new telecommunication services-WiBro and HSDPA. We began with a randomly selected 800 interviewees, and reduced them to 432 because other remaining ones were related to the people who did not show strong intention for subscription to new telecommunications services. For each combination of content and handset, number of responses which favored WiBro and HSDPA were counted, respectively. It was assumed that interviewee favors a specific service when expected utility is greater than that of competing service(s). Then, the market share of each service was calculated by normalizing the total number of responses which preferred each service. Holistic evaluation of new and unfamiliar service is a tough challenge for survey respondents. We have developed a simple and easy method to assess individual level preference by estimating weight of each attribute. Swing method was applied for this purpose. We believe that estimating individual level preference will be quite flexibly used to predict market performance of new services in many different business environments.