• 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 the Korean Data and Information Science Society
• /
• v.19 no.4
• /
• pp.1233-1246
• /
• 2008

This study examines characteristics of people who are experiencing severe hair loss problems. We focus on how these characteristics are related to their psychological problems, hair loss treatments, wig-wearing practices, and life styles. We gathered survey data from people who visited wig shops for their hair loss problems. The study shows that men and women have different characteristics in every aspect we consider.

• Korean Management Science Review
• /
• v.28 no.3
• /
• pp.73-82
• /
• 2011

Basel II advanced measurement approaches for operational risk need to estimate the frequency and severity distribution of operational losses. Due to lack of internal loss data, the estimation is impossible in many cases and so external loss data might be used by scaling on asset or gross income. To get around lack of loss data, scenario analysis combined with loss distribution approach can be useful in calculating the capital charge of operational risk. However, scenario based loss distribution approach requires much time and effort. Instead we may apply the analytic hierarchy process to measure operational risk of financial institutions. The analytic hierarchy process combined with loss distribution approach is to estimate the capital charge of operational risk in other areas based on the operational VaR in an area with sufficient loss data. AHP provides a tool for timely measurement of operational risk in this rapidly changing global environment.

• Journal of the Korean Society for Precision Engineering
• /
• v.17 no.2
• /
• pp.35-42
• /
• 2000

In recent years the recording density of hard disks has increased significantly largely due to the decreasing flying height. As a result of decreased flying height, the reliability issue become more critical. In this work the relationship between hard disk surface damage and data loss was investigated by using an actual hard disk drive. The purpose of this research was to identify the key factor which leads to data loss. It was shown that data loss is directly related to the physical damage of the Co-magnetic layer and there was no data loss when only carbon protective coating was damaged by the diamond tip.

• Journal of the Korean Data and Information Science Society
• /
• v.15 no.4
• /
• pp.899-910
• /
• 2004

In this paper we have proposed a new loss function, namely, non-linear exponential(NLINEX) loss function, which is quite asymmetric in nature. We obtained the Bayes estimator under exponential(LINEX) and squared error(SE) loss functions. Moreover, a numerical comparison among the Bayes estimators of power function distribution under SE, LINEX, and NLINEX loss function have been made.

• International journal of advanced smart convergence
• /
• v.12 no.2
• /
• pp.173-181
• /
• 2023

Data collection is an essential element in the construction and operation of a smart factory. The quality of data collection is greatly influenced by network conditions, and existing wireless network systems for IoT inevitably lose data due to wireless signal strength. This data loss has contributed to increased system instability due to misinformation based on incorrect data. In this study, I designed a distributed MQTT IoT smart sensor and gateway structure that supports wireless multicasting for smooth sensor data collection. Through this, it was possible to derive significant results in the service latency and data loss rate of packets even in a wireless environment, unlike the MQTT QoS-based system. Therefore, through this study, it will be possible to implement a data collection management system optimized for the domestic smart factory manufacturing environment that can prevent data loss and delay due to abnormal data generation and minimize the input of management personnel.

• Journal of Distribution Science
• /
• v.15 no.11
• /
• pp.21-29
• /
• 2017

Purpose - Relationship between farm and county losses determines whether the county program provides too little, too much, or similar amount of assistance relative to the loss on an individual farm. A review of the literature finds limited analysis of the determinants of this relationship. This paper conducts such an analysis using farm-level yield data. Research design, data, and methodology - Farm-level yield data from Illinois and Kansas farm business management associations are used for to calculate the correlation between farm and county loss and the share of farm loss systemic with county loss, and also for the regression analysis. Results - Average share of farm loss systemic with the county loss lies between 42% and 68%. The correlation between farm and county yield/revenue deviation from expected value is statistically significant in all four models. The coefficient is positive, implying the higher the correlation, the larger the share of farm loss that is systemic with the county loss. Conclusions - The findings of this study are consistent with the existing literature which argues that county variability may not be closely associated with farm variability. The findings of this study thus raise questions about the efficacy of area yield and revenue insurance products in helping farmers manage their risk.

• Journal of Periodontal and Implant Science
• /
• v.52 no.6
• /
• pp.509-521
• /
• 2022

Purpose: Systemic health has a profound effect on dental treatment. The aim of this study was to evaluate peri-implant bone loss and health screening data to discover factors that may influence peri-implant diseases. Methods: This study analyzed the panoramic X-rays of patients undergoing health screenings at the Health Promotion Center at Seoul St. Mary's Hospital in 2018, to investigate the relationship between laboratory test results and dental data. The patients' physical data, such as height, weight, blood pressure, hematological and urine analysis data, smoking habits, number of remaining teeth, alveolar bone level, number of implants, and degree of bone loss around the implant, were analyzed for correlations. Their associations with glycated hemoglobin, glucose, blood urea nitrogen (BUN), creatinine, and severity of periodontitis were evaluated using univariate and multivariate regression analysis. Results: In total, 2,264 patients opted in for dental health examinations, of whom 752 (33.2%) had undergone dental implant treatment. These 752 patients had a total of 2,658 implants, and 129 (17.1%) had 1 or more implants with peri-implant bone loss of 2 mm or more. The number of these implants was 204 (7%). Body mass index and smoking were not correlated with peri-implant bone loss. Stepwise multivariate regression analysis revealed that the severity of periodontal bone loss (moderate bone loss: odds ratio [OR], 3.154; 95% confidence interval [CI], 1.175-8.475 and severe bone loss: OR, 7.751; 95% CI, 3.003-20) and BUN (OR, 1.082; 95% CI, 1.027-1.141) showed statistically significant predictive value. The severity of periodontitis showed greater predictive value than the biochemical parameters of blood glucose, renal function, and liver function. Conclusions: The results of this study showed that periodontal bone loss was a predictor of peri-implant bone loss, suggesting that periodontal disease should be controlled before dental treatment. Diligent maintenance care is recommended for patients with moderate to severe periodontal bone loss.

• Journal of the Korean Data and Information Science Society
• /
• v.14 no.3
• /
• pp.631-639
• /
• 2003

Bayes estimates of reliability for the stress-strength system are obtained with respect to LINEX loss function. A reference prior distribution of the reliability is derived and Bayes estimates of the reliability are also obtained. These Bayes estimates are compared with corresponding estimates under squared-error loss function.

• International conference on construction engineering and project management
• /
• 2022.06a
• /
• pp.808-813
• /
• 2022

This study presents the problems of data imbalance, varying difficulties across target objects, and small objects in construction object segmentation for far-field monitoring and utilize compound loss functions to address it. Construction site scenes of assembling scaffolds were analyzed to test the effectiveness of compound loss functions for five construction object classes---workers, hardhats, harnesses, straps, hooks. The challenging problem was mitigated by employing a focal and Jaccard loss terms in the original loss function of LinkNet segmentation model. The findings indicates the importance of the loss function design for model performance on construction site scenes for far-field monitoring.