• Journal of the Korean Data and Information Science Society
• /
• v.25 no.3
• /
• pp.685-696
• /
• 2014

In this paper, we develop Bayesian inference of the finite population mean with the assumption of posterior linearity rather than normality of the superpopulation in the presence of auxiliary information under the balanced loss function. We compare the performance of the optimal Bayes estimator under the balanced loss function with ones of the classical ratio estimator and the usual Bayes estimator in terms of the posterior expected losses, risks and Bayes risks.

• Journal of Korean Institute of Industrial Engineers
• /
• v.31 no.2
• /
• pp.164-172
• /
• 2005

A loss function approach to a multiresponse problem is considered, when process parameters are regarded as random variables. The variation of each response may be amplified through so called propagation of error (POE), which is defined as the standard deviation of the transmitted variability in the response as a function of process parameters. The forms of POE for each response and for a pair of responses are proposed and they are reflected in our loss function approach to determine the optimal condition. The proposed method is illustrated using a polymer case. The result is compared with the case where parameter fluctuation is not considered.

• Journal of Electrical Engineering and Technology
• /
• v.8 no.3
• /
• pp.428-435
• /
• 2013

Flexible AC Transmission System (FACTS) application study on enhancing the flexibility of AC power system has continued to make progress. A thyristor-controlled series capacitor (TCSC) is a useful FACTS device that can control the power flow by adjusting line impedances and minimize the loss of power flow and voltage drop in a transmission system by adjusting line impedances. Reduced power flow loss leads to increased loadability, low system loss, and improved stability of the power system. This study proposes the optimal location and compensation rate method for TCSCs, by considering both the power system loss and voltage drop of transmission systems. The proposed method applies a multi-objective function consisting of a minimizing function for power flow loss and voltage drop. The effectiveness of the proposed method is demonstrated using IEEE 14- and a 30-bus system.

• Journal of Korea Multimedia Society
• /
• v.22 no.8
• /
• pp.844-851
• /
• 2019

In this paper, we propose a loss function in CNN that introduces inter-class amplitudes to increase inter-class loss and reduce intra-class loss to increase of face recognition performance. This loss function increases the distance between the classes and decreases the distance in the class, thereby improving the performance of the face recognition finally. It is confirmed that the accuracy of face recognition for visible light image of proposed loss function is 99.62%, which is better than other loss functions. We also applied it to face recognition of visible and near-infrared complex images to obtain satisfactory results of 99.76%.

• The Korean Journal of Applied Statistics
• /
• v.27 no.6
• /
• pp.1017-1028
• /
• 2014

Well-known Bayes and empirical Bayes estimators have a disadvantage in respecting to overshink the parameter estimator error; therefore, a constrained Bayes estimator is suggested by matching the first two moments. Also traditional loss function such as mean square error loss function only considers the precision of estimation and to consider both precision and goodness of fit, balanced loss function is suggested. With these reasons, constrained Bayes estimators under balanced loss function is recommended for non-life insurance pricing.; however, most studies focus on the performance of estimation since Bayes risk of newly suggested estimators such as constrained Bayes and constrained empirical Bayes estimators under specific loss function is difficult to derive. This study compares the Bayes risk of several Bayes estimators under two different loss functions for estimating the risk in the auto insurance business and indicates the effectiveness of the newly suggested Bayes estimators with regards to Bayes risk perspective through auto insurance real data analysis.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.24 no.1
• /
• pp.49-54
• /
• 2024

Data imbalance is a common issue encountered in classification problems, stemming from a significant disparity in the number of samples between classes within the dataset. Such data imbalance typically leads to problems in classification models, including overfitting, underfitting, and misinterpretation of performance metrics. Methods to address this issue include resampling, augmentation, regularization techniques, and adjustment of loss functions. In this paper, we focus on loss function adjustment, particularly comparing the performance of various configurations of loss functions (Cross Entropy, Balanced Cross Entropy, two settings of Focal Loss: 𝛼 = 1 and 𝛼 = Balanced, Asymmetric Loss) on Multi-Class black-box video data with imbalance issues. The comparison is conducted using the I3D, and R3D_18 models.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.41 no.1
• /
• pp.110-117
• /
• 2018

Machines and facilities are physically or chemically degenerated by continuous usage. One of the results of this degeneration is the process mean shift. By the result of degeneration, non-conforming products and malfunction of machine occur. Therefore a periodic preventive resetting the process is necessary. This type of preventive action is called 'preventive maintenance policy.' Preventive maintenance presupposes that the preventive (resetting the process) cost is smaller than the cost of failure caused by the malfunction of machine. The process mean shift problem is a field of preventive maintenance. This field deals the interrelationship between the quality cost and the process resetting cost before machine breaks down. Quality cost is the sum of the non-conforming item cost and quality loss cost. Quality loss cost is due to the deviation between the quality characteristics from the target value. Under the process mean shift, the quality cost is increasing continuously whereas the process resetting cost is constant value. The objective function is total costs per unit wear, the decision variables are the wear limit (resetting period) and the initial process mean. Comparing the previous studies, we set the process variance as an increasing concave function and set the quality loss function as Cpm+ simultaneously. In the Cpm+, loss function has different cost coefficients according to the direction of the quality characteristics from target value. A numerical example is presented.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.7
• /
• pp.2339-2355
• /
• 2021

In this paper, a joint probabilistic data association algorithm based on loss function (LJPDA) is proposed so that the computation load and accuracy of the multi-target tracking algorithm can be guaranteed simultaneously. Firstly, data association is divided in to three cases based on the relationship among validation gates and the number of measurements in the overlapping area for validation gates. Also the contribution coefficient is employed for evaluating the contribution of a measurement to a target, and the loss function, which reflects the cost of the new proposed data association algorithm, is defined. Moreover, the equation set of optimal contribution coefficient is given by minimizing the loss function, and the optimal contribution coefficient can be attained by using the Newton-Raphson method. In this way, the weighted value of each target can be achieved, and the data association among measurements and tracks can be realized. Finally, we compare performances of LJPDA proposed and joint probabilistic data association (JPDA) algorithm via numerical simulations, and much attention is paid on real-time performance and estimation error. Theoretical analysis and experimental results reveal that the LJPDA algorithm proposed exhibits small estimation error and low computation complexity.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.41 no.4
• /
• pp.203-212
• /
• 2018

Quality requirements of manufactured products or parts are given in the form of specification limits on the quality characteristics of individual units. If a product is to meet the customer's fitness for use criteria, it should be produced by a process which is stable or repeatable. In other words, it must be capable of operating with little variability around the target value or nominal value of the product's quality characteristic. In order to maintain and improve product quality, we need to apply statistical process control techniques such as histogram, check sheet, Pareto chart, cause and effect diagram, or control charts. Among those techniques, the most important one is control charting. The cumulative sum (CUSUM) control charts have been used in statistical process control (SPC) in industries for monitoring process shifts and supporting online measurement. The objective of this research is to apply Taguchi's quality loss function concept to cost based CUSUM control chart design. In this study, a modified quality loss function was developed to reflect quality loss situation where general quadratic loss curve is not appropriate. This research also provided a methodology for the design of CUSUM charts using Taguchi quality loss function concept based on the minimum cost per hour criterion. The new model differs from previous models in that the model assumes that quality loss is incurred even in the incontrol period. This model was compared with other cost based CUSUM models by Wu and Goel, According to numerical sensitivity analysis, the proposed model results in longer average run length in in-control period compared to the other two models.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2003.09b
• /
• pp.45-48
• /
• 2003

We propose some properties of Bayesian fuzzy hypotheses testing by revision for prior possibility distribution and posterior possibility distribution using weighted fuzzy hypotheses H$\sub$0/($\theta$) versus H$_1$($\theta$) on $\theta$ with loss function.