• Journal of KIISE:Computing Practices and Letters
• /
• v.11 no.2
• /
• pp.112-123
• /
• 2005

The latest E-commerce sites provide personalized services to maximize user satisfaction for Internet user The collaborative filtering is an algorithm for personalized item real-time recommendation. Various supplementary methods are provided for improving the accuracy of prediction and performance. It is important to consider these two things simultaneously to implement a useful recommendation system. However, established studies on collaborative filtering technique deal only with the matter of accuracy improvement and overlook the matter of performance. This study considers representative attribute-neighborhood, recommendation textile set, and similarity grouping that are expected to improve performance to the recommendation agent system. Ultimately, this paper suggests empirical applications to verify the adequacy and the validity on this system with the development of Fashion Design Recommendation Agent System (FDRAS ).

• Proceedings of the Korean Society of Near Infrared Spectroscopy Conference
• /
• 2001.06a
• /
• pp.1141-1141
• /
• 2001

The aim of the study was to evaluate the performance of 3 calibration methods, modified partial least squares (MPLS), local PLS (LOCAL) and artificial neural network (ANN) on the prediction of chemical composition of forages, using a large NIR database. The study used forage samples (n=25,977) from Australia, Europe (Belgium, Germany, Italy and Sweden) and North America (Canada and U.S.A) with information relative to moisture, crude protein and neutral detergent fibre content. The spectra of the samples were collected with 10 different Foss NIR Systems instruments, which were either standardized or not standardized to one master instrument. The spectra were trimmed to a wavelength range between 1100 and 2498 nm. Two data sets, one standardized (IVAL) and the other not standardized (SVAL) were used as independent validation sets, but 10% of both sets were omitted and kept for later expansion of the calibration database. The remaining samples were combined into one database (n=21,696), which was split into 75% calibration (CALBASE) and 25% validation (VALBASE). The chemical components in the 3 validation data sets were predicted with each model derived from CALBASE using the calibration database before and after it was expanded with 10% of the samples from IVAL and SVAL data sets. Calibration performance was evaluated using standard error of prediction corrected for bias (SEP(C)), bias, slope and R2. None of the models appeared to be consistently better across all validation sets. VALBASE was predicted well by all models, with smaller SEP(C) and bias values than for IVAL and SVAL. This was not surprising as VALBASE was selected from the calibration database and it had a sample population similar to CALBASE, whereas IVAL and SVAL were completely independent validation sets. In most cases, Local and ANN models, but not modified PLS, showed considerable improvement in the prediction of IVAL and SVAL after the calibration database had been expanded with the 10% samples of IVAL and SVAL reserved for calibration expansion. The effects of sample processing, instrument standardization and differences in reference procedure were partially confounded in the validation sets, so it was not possible to determine which factors were most important. Further work on the development of large databases must address the problems of standardization of instruments, harmonization and standardization of laboratory procedures and even more importantly, the definition of the database population.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.52 no.10
• /
• pp.33-46
• /
• 2015

In this paper, we predicts the analog and digital circuit performance of FinFETs that are scaled down following the ITRS(International technology roadmap for semiconductors). For accurate prediction of the circuit performance of scaled down devices, accurate parasitic resistance and capacitance analytical models are developed and their accuracies are within 2 % compared to 3D TCAD simulation results. The parasitic capacitance models are developed using conformal mapping, and the parasitic resistance models are enhanced to include the fin extension length($L_{ext}$) with respect to the default parasitic resistance model of BSIM-CMG. A new algorithm is developed to fit the DC characteristics of BSIM-CMG to the reference DC data. The proposed capacitance and resistance models are implemented inside BSIM-CMG to replace the default parasitic model, and SPICE simulations are performed to predict circuit performances such as $f_T$, $f_{MAX}$, ring oscillators and common source amplifier. Using the proposed parasitic capacitance and resistance model, the device and circuit performances are quantitatively predicted down to 5 nm FinFET transistors. As the FinFET technology scales, due to the improvement in both DC characteristics and the parasitic elements, the circuit performance will improve.

• Journal of Platform Technology
• /
• v.11 no.3
• /
• pp.45-55
• /
• 2023

This paper presents a model that supports decision-makers in the Korean film industry to maximize the success of online movies. To achieve this, we collected historical box office movies and clustered them into types to propose a model predicting each type's online box office performance. We considered various features to identify factors contributing to movie success and reduced feature dimensionality for computational efficiency. We systematically classified the movies into types and predicted each type's online box office performance while analyzing the contributing factors. We used automated machine learning (AutoML) techniques to automatically propose and select machine learning algorithms optimized for the problem, allowing for easy experimentation and selection of multiple algorithms. This approach is expected to provide a foundation for informed decision-making and contribute to better performance in the film industry.

• Journal of Environmental Impact Assessment
• /
• v.31 no.2
• /
• pp.83-94
• /
• 2022

In this study, DNN-based models were learned using air quality determination data for 2017, 2019, and 2020 provided by the National Measurement Network (Air Korea), and this models evaluated using data from 2016 and 2018. Based on Pearson correlation coefficient 0.2, four items (SO₂, CO, NO₂, PM₁₀) were initially modeled as independent variables. In order to improve the accuracy of prediction, monthly independent modeling was carried out. The error was calculated by RMSE (Root Mean Square Error) method, and the initial model of RMSE was 5.78, which was about 46% betterthan the national moving average modelresult (10.77). In addition, the performance improvement of the independent monthly model was observed in months other than November compared to the initial model. Therefore, this study confirms that DNN modeling was effective in predicting PM_2.5 concentrations based on air pollutants concentrations, and that the learning performance of the model could be improved by selecting additional independent variables.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.12 no.6
• /
• pp.629-635
• /
• 2019

In Deep Learning models derivative is implemented by error back-propagation which enables the model to learn the error and update parameters. It can find the global (or local) optimal points of parameters even in the complex models taking advantage of a huge improvement in computing power. However, deliberately generated data points can 'fool' models and degrade the performance such as prediction accuracy. Not only these adversarial examples reduce the performance but also these examples are not easily detectable with human's eyes. In this work, we propose the method to detect adversarial datasets with random noise addition. We exploit the fact that when random noise is added, prediction accuracy of non-adversarial dataset remains almost unchanged, but that of adversarial dataset changes. We set attack methods (FGSM, Saliency Map) and noise level (0-19 with max pixel value 255) as independent variables and difference of prediction accuracy when noise was added as dependent variable in a simulation experiment. We have succeeded in extracting the threshold that separates non-adversarial and adversarial dataset. We detected the adversarial dataset using this threshold.

• Geomechanics and Engineering
• /
• v.28 no.4
• /
• pp.385-396
• /
• 2022

The accurate determination of the subgrade reaction modulus (K_s) of soil is an important factor for geotechnical engineers. This study estimated the K_s of soft soil improved with floating deep cement mixing (DCM) columns. A novel prediction model was developed that emphasizes the accuracy of identifying the most significant parameters of K_s. Several multi-layer perceptron (MLP) models that were trained using the Levenberg Marquardt (LM) backpropagation method were developed to estimate K_s. The models were trained using a reliable database containing the results of 36 physical modelling tests. The input parameters were the undrained shear strength of the DCM columns, undrained shear strength of soft soil, area improvement ratio and length-to-diameter ratio of the DCM columns. Grey wolf optimization (GWO) was coupled with the MLPs to improve the performance indices of the MLPs. Sensitivity tests were carried out to determine the importance of the input parameters for prediction of K_s. The results showed that both the MLP-LM and MLP-GWO methods showed high ability to predict K_s. However, it was shown that MLP-GWO (R = 0.9917, MSE = 0.28 (MN/m²/m)) performed better than MLP-LM (R =0.9126, MSE =6.1916 (MN/m²/m)). This proves the greater reliability of the proposed hybrid model of MLP-GWO in approximating the subgrade reaction modulus of soft soil improved with floating DCM columns. The results revealed that the undrained shear strength of the soil was the most effective factor for estimation of K_s.

• Structural Engineering and Mechanics
• /
• v.91 no.6
• /
• pp.643-656
• /
• 2024

The prediction of the moment rotation behavior of semi-rigid connections has been the subject of extensive research. However, to improve the accuracy of these predictions, there is a growing interest in employing machine learning algorithms. This paper investigates the effectiveness of using Multi-gene genetic programming (MGGP) to predict the moment-rotation behavior of flush-end plate connections compared to that of artificial neural networks (ANN) and previous studies. It aims to automate the process of determining the most suitable equations to accurately describe the behavior of these types of connections. Experimental data was used to train ANN and MGGP. The performance of the models was assessed by comparing the values of coefficient of determination (R²), maximum absolute error (MAE), and root-mean-square error (RMSE). The results showed that MGGP produced more accurate, reliable, and general predictions compared to ANN and previous studies with an R² exceeding 0.99, an RMSE of 6.97, and an MAE of 38.68, highlighting its advantages over other models. The use of MGGP can lead to better modeling and more precise predictions in structural design. Additionally, an experimentally-based regression analysis was conducted to obtain the rotational capacity of FECs. A new equation was proposed and compared to previous ones, showing significant improvement in accuracy with an R² score of 0.738, an RMSE of 0.014, and an MAE of 0.024.

• Proceedings of the Korean Geotechical Society Conference
• /
• 1998.05a
• /
• pp.153-204
• /
• 1998

Prediction of lining loads due to tunnelling is one of the major issues to be addressed in the design of a tunnel. The objective of this study is to investigate rational and realistic design loads on tunnel linings. factors influencing the lining load are summarized and discussed. The instruments for measuring the lining loads are reviewed and discussed because field measurements are often necessary to verify the design methods. Tunnel construction in the City of Edmonton has been very active for storm and sanitary purposes. Since the early 1970's, the city has also been developing an underground Light Rail Transit system. The load measurements obtained from these tunnels are compared with the results from the existing design methods. However, none of the existing methods are totally satisfactory, Therefore, there is some room for improvement in the prediction of lining loads. The convergence-confinement method is reviewed and applied to a case history of a tunnel in Edmonton. The convergence curves are obtained from 2-D finite element analyses using three different material models and theoretical equations. The limitation of the convergence-confinement method is discussed by comparing these curves with the field measurements. Three-dimensional finite element analyses are performed to gain a better understanding of stress and displacement behaviour near the tunnel face. An improved design method is proposed based on the review of existing design methods and the performance of numerical analyses. A specific method or combination of two different methods is suggested for the estimation of lining loads for different conditions of tunnelling. A method to determine the stress reduction factor is described. Typical values of dimensionless load factors nD/H for tunnels in Edmonton are obtained from parametric analyses. Finally, the loads calculated using the proposed method are compared with field measurements collected from various tunnels in terms of soil types and construction methods to verify the method. The proposed method gives a reasonable approximation of the lining loads. The proposed method is recommended as an approximate guideline for the design of tunnels, but the results should be confirmed by field measurements due to the uncertainties of the ground and lining properties and the construction procedures, This is the reason that in-situ monitoring should be an integral part of the design procedure.

• International Journal of Concrete Structures and Materials
• /
• v.9 no.2
• /
• pp.159-172
• /
• 2015

Estimation of the residual strength of corroded reinforced concrete beams has been studied from experimental and theoretical perspectives. The former is arduous as it involves casting beams of various sizes, which are then subjected to various degrees of corrosion damage. The latter are static; hence cannot be generalized as new coefficients need to be re-generated for new cases. This calls for dynamic models that are adaptive to new cases and offer efficient generalization capability. Computational intelligence techniques have been applied in Construction Engineering modeling problems. However, these techniques have not been adequately applied to the problem addressed in this paper. This study extends the empirical model proposed by Azad et al. (Mag Concr Res 62(6):405-414, 2010), which considered all the adverse effects of corrosion on steel. We proposed four artificial neural networks (ANN) models to predict the residual flexural strength of corroded RC beams using the same data from Azad et al. (2010). We employed two modes of prediction: through the correction factor ($C_f$) and through the residual strength ($M_{res}$). For each mode, we studied the effect of fixed and random data stratification on the performance of the models. The results of the ANN models were found to be in good agreement with experimental values. When compared with the results of Azad et al. (2010), the ANN model with randomized data stratification gave a $C_f$-based prediction with up to 49 % improvement in correlation coefficient and 92 % error reduction. This confirms the reliability of ANN over the empirical models.