• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.41 no.2
• /
• pp.174-183
• /
• 2018

In the industrial fields, the process capability index has been using to evaluate the variation of quality in the process. The traditional process capability indices such as $C_p$, $C_{pk}$, $C_{pm}$ and $C^+_{pm}$ have been applied in the industrial fields. These traditional process capability indices are mainly applied in the univariate analysis. However, the main streams in the recent industry are the multivariate manufacturing process and the multiple quality characteristics are corrected each other. Therefore, the multivariate statistical method should be used in the process capability analysis. The multivariate process indices need to be enhanced with more useful information and extensive application in the recent industrial fields. Hence, the purpose of the study is to develop a more effective multivariate process index ($MC_{pI}$) using the multivariate inverted normal loss function. The multivariate inverted normal loss function has the flexibility for the any type of the symmetrical and asymmetrical loss functions as well as the economic information. Especially, the proposed modeling method for the multivariate inverted normal loss function (MINLF) and the expected loss from MINLF in this paper can be applied to the any type of the symmetrical and asymmetrical loss functions. And this modeling method can be easily expanded from a bivariate case to a multivariate case.

• Smart Structures and Systems
• /
• v.29 no.1
• /
• pp.153-166
• /
• 2022

Identifying fine cracks in steel bridge facilities is a challenging task of structural health monitoring (SHM). This study proposed an end-to-end crack image segmentation framework based on a one-step Convolutional Neural Network (CNN) for pixel-level object recognition with high accuracy. To particularly address the challenges arising from small object detection in complex background, efforts were made in loss function selection aiming at sample imbalance and module modification in order to improve the generalization ability on complicated images. Specifically, loss functions were compared among alternatives including the Binary Cross Entropy (BCE), Focal, Tversky and Dice loss, with the last three specialized for biased sample distribution. Structural modifications with dilated convolution, Spatial Pyramid Pooling (SPP) and Feature Pyramid Network (FPN) were also performed to form a new backbone termed CrackDet. Models of various loss functions and feature extraction modules were trained on crack images and tested on full-scale images collected on steel box girders. The CNN model incorporated the classic U-Net as its backbone, and Dice loss as its loss function achieved the highest mean Intersection-over-Union (mIoU) of 0.7571 on full-scale pictures. In contrast, the best performance on cropped crack images was achieved by integrating CrackDet with Dice loss at a mIoU of 0.7670.

• Communications for Statistical Applications and Methods
• /
• v.7 no.3
• /
• pp.767-772
• /
• 2000

Zellner(1994) introduced the notion of a balanced loss function in the context of a general liner model to reflect both goodness of fit and precision of estimation. We study the perspective of unifying a variety of results both frequentist and Bayesian from Poisson distributions. We show that frequentist and Bayesian results for balanced loss follow from and also imply related results for quadratic loss functions reflecting only precision of estimation. Several examples are given for Poisson distribution.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.19 no.40
• /
• pp.233-240
• /
• 1996

A fixed amount compensator is proposed for a process with a linear tool wear function. A Cost model is constructed which involve process adjustment cost and quality loss. Symmetric and asymmetric quadratic functions of the deviation of a quality measurement from the nominal target value are considered as the quality loss functions. Methods of finding optimal values of initial setting and compensation limit are presented and a numerical example is given.

• Journal of Korea Multimedia Society
• /
• v.24 no.10
• /
• pp.1380-1390
• /
• 2021

The video face recognition (FR) is one of the most popular researches in the field of computer vision due to a variety of applications. In particular, research using the attention mechanism is being actively conducted. In video face recognition, attention represents where to focus on by using the input value of the whole or a specific region, or which frame to focus on when there are many frames. In this paper, we propose a novel attention based deep learning method. Main novelties of our method are (1) the use of combining two loss functions, namely weighted Softmax loss function and a Triplet loss function and (2) the feasibility of end-to-end learning which includes the feature embedding network and attention weight computation. The feature embedding network has a positive effect on the attention weight computation by using combined loss function and end-to-end learning. To demonstrate the effectiveness of our proposed method, extensive and comparative experiments have been carried out to evaluate our method on IJB-A dataset with their standard evaluation protocols. Our proposed method represented better or comparable recognition rate compared to other state-of-the-art video FR methods.

• Journal of the Earthquake Engineering Society of Korea
• /
• v.23 no.2
• /
• pp.131-140
• /
• 2019

Diagonally reinforced concrete coupling beams (DRCBs) have been widely adopted in reinforced concrete (RC) bearing wall systems. DRCBs are known to act as a fuse element dissipating most of seismic energies imparted to the bearing wall systems during earthquakes. Despite such importance of DRCBs, the damage estimation of such components and the corresponding consequences within the knowledge of performance based seismic design framework is not well understood. In this paper, drift-based fragility functions are developed for in-plane loaded DRCBs. Fragility functions are developed to predict the damage and to decide the repair method required for DRCBs subjected to earthquake loading. Thirty-seven experimental results are collected from seventeen published literatures for this effort. Drift-based fragility functions are developed for four damage states of DRCBs subjected to cyclic and monotonic loading associated with minor cracking, severe cracking, onset of strength loss, and significant strength loss. Damage states are defined in a consistent manner. Cumulative distribution functions are fit to the empirical data and evaluated using standard statistical methods.

• Journal of Power Electronics
• /
• v.14 no.2
• /
• pp.351-361
• /
• 2014

The discrete torque generation mechanism and inherently nonlinear magnetic characterization of switched reluctance motors lead to unacceptable torque ripples and limit the application of these motors. In this study, a phase current profiling technique and torque sharing function are proposed in consideration of magnetic saturation effects and by minimizing power loss in the commutation area between the adjacent phases. Constant torque trajectories are considered in incoming and outgoing phase current planes based on nonlinear T-i-theta curves obtained from experimental measurements. Optimum points on constant torque trajectories are selected by improving drive efficiency and minimizing copper loss in each rotor position. A novel analytic invertible function is introduced to express phase torque based on rotor position and its corresponding phase current. The optimization problem is solved by the proposed torque function, and optimum torque sharing functions are derived. A modification method is also introduced to enhance the torque ripple-free region based on simple logic rules. Compared with conventional torque sharing functions, the resultant reference current from the proposed method has less peak and effective values and exhibits lower copper loss. Experimental and simulation results from a four-phase 4 KW 8/6 SRM validate the effectiveness of the proposed method.

• Proceedings of the Korean Institute of Building Construction Conference
• /
• 2019.05a
• /
• pp.125-126
• /
• 2019

This study aims to recognize damage indicators of typhoon and to develop damage function's indicators, using information derived from the actual loss of typhoon Maemi. As typhoons engender significant financial damage all over the world, governments and insurance companies, local or global, develop hurricane risk assessment models and use it in quantifying, avoiding, mitigating, or transferring the risks. For the reason, it is crucial to understand the importance of the risk assessment model for typhoons, and the importance of reflecting local vulnerabilities for more advanced evaluation. Although much previous research on the economic losses associated with natural disasters has identified the risk indicators that are indispensable, more comprehensive research addressing the relationship between vulnerability and economic loss are still called for. Hence this study utilizes and analyzes the actual loss record of the typhoon Maemi provided by insurance companies to fill such gaps. In this study, natural disaster indicators and basic building information indicators are used in order to generate the vulnerability functions; and the results and indicators suggest a practical approach to create the vulnerability functions for insurance companies and administrative tasks, while reflecting the financial loss and local vulnerability of the actual buildings.

• Journal of the Korean Statistical Society
• /
• v.31 no.4
• /
• pp.415-431
• /
• 2002

A unified framework for statistical inference for the mean of the normal distribution to derive point estimates, confidence intervals and statistical tests is proposed. This optimal design is justified after investigating the basic information and requirements that are possible and impossible to control when specifying practical and statistical requirements. Point estimation is only credible when viewed in the larger context of interval estimation, since the information required for optimal point estimation is unspecifiable. Triple sampling is proposed and justified as a reasonable sampling vehicle to achieve the specifiable requirements within the unified framework.

• Journal of the Korean Statistical Society
• /
• v.20 no.2
• /
• pp.202-207
• /
• 1991

In Le Cam's earlier work on superefficiency, it is proved that if an estimate is superefficient at a given paramter value $\theta$$\_$0/, then there must exist an infinite sequence {$\theta$$\_$n/}) of values(conversing to $\theta$$\_$0/) at which this estimate is worse than M. L. E. for certain classes of loss functions. For one-dimensional cases, these classes of lass functions include squared error loss. However. for multi-dimensional cases, they do not. This note is to give an example where a superefficiest estimator of a multi-dimensional parameter is not inferior to M. L. E. along any sequence ($\theta$$\_$n/) converging to the point of superefficiency with respect to the squared error loss.