• Proceedings of the Korean Society for Quality Management Conference
• /
• 2004.04a
• /
• pp.655-673
• /
• 2004

This paper is concerned with analysing the bankruptcy prediction power of three models: Multivariate Discriminant Analysis(MDA ), Logit Analysis, Neural Network. The after-crisis bankrupted companies were limited to the research data and the listed companies belonging to manufacturing industry was limited to the research data so as to improve prediction accuracy and validity of the model. In order to assure meaningful bankruptcy prediction, training data and testing data were not extracted within the corresponding period. The result is that prediction accuracy of neural network model is more excellent than that of logit analysis and MDA model when considering that execution of testing data was followed by execution of training data.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.16 no.1
• /
• pp.1-8
• /
• 2021

Detecting Out-of-Distribution (OOD) data is fundamentally required when Deep Neural Network (DNN) is applied to real-world AI such as autonomous driving. However, modern DNNs are quite vulnerable to the over-confidence problem even if the test data are far away from the trained data distribution. To solve the problem, this paper proposes a novel Adversarial-Mixup training method to let the DNN model be more robust by detecting OOD data effectively. Experimental results show that the proposed Adversarial-Mixup method improves the overall performance of OOD detection by 78% comparing with the State-of-the-Art methods. Furthermore, we show that the proposed method can alleviate the over-confidence problem by reducing the confidence score of OOD data than the previous methods, resulting in more reliable and robust DNNs.

• Journal of information and communication convergence engineering
• /
• v.22 no.3
• /
• pp.249-255
• /
• 2024

Advancements in deep learning have enhanced vision-based aggregate analysis. However, further development and studies have encountered challenges, particularly in acquiring large-scale datasets. Data collection is costly and time-consuming, posing a significant challenge in acquiring large datasets required for training neural networks. To address this issue, this study introduces a simulation that efficiently generates the necessary data and labels for training neural networks. We utilized a genetic algorithm (GA) to create optimized lists of aggregates based on the specified values of weight and particle size distribution for the aggregate sample. This enabled sample data collection without conducting sieving tests. Our evaluation of the proposed simulation and GA methodology revealed errors of 1.3% and 2.7 g for aggregate size distribution and weight, respectively. Furthermore, we assessed a segmentation model trained with data from the simulation, achieving a promising preliminary F1 score of 78.18 on the actual aggregate image.

• Journal of the Korean Data and Information Science Society
• /
• v.25 no.3
• /
• pp.555-565
• /
• 2014

The purpose of this study was to analyze sports injuries which occurred in Korea National Team during official training period. All sports injuries are recorded on injury report form by physicians, medical staffs and athletic trainer, and only acute and recurred injuries were analyzed. Total 3,421 injuries were reported, and 1,560 injuries were newly incurred and 1,861 injuries were recurrent with previous history. The frequency of new injuries in male and female athletes was highest in boxing (n=130, 14.5%) and hockey (n=75, 11.3%) respectively. The frequency of recurred injuries in male and female athletes was highest in wrestling (n=147, 14.8%) and fencing (n=103, 11.9%) respectively. Our data provides incidence rates, characteristics of acute and recurrent sports injuries during official training period and thus these results could provide relevant information for the sports injury prevention at Korea National Team player.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.05a
• /
• pp.147-149
• /
• 2022

This paper proposes a state machine design methodology in which an interactive robot that mimics discrete trial training (DTT protocol) can support social interaction training for children with autism. The robot applied to social interaction training uses the response to the provided training stimulus as a quantitative indicator by processing the data received from the sensors measuring the behavioral response of the child. In this process, the state machine is used as information that classifies the state of the acquired data and provides the subsequent stimulus for DTT protocol. Through the joint attentional training, it can be used as evidence-based treatment information by quantitatively classifying the data on the number of sustainable and DTT protocol and the child's response, as well as the current reaction status of the child to the observer performing remote monitoring. At the same time, it was confirmed that it is possible to properly respond to misrecognition situations.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.17 no.4
• /
• pp.213-220
• /
• 2024

This paper proposes an optimization strategy for performing Federated Learning between devices and edge clouds using WebAssembly (WASM). The proposed strategy aims to maximize efficiency by conducting partial training on devices and the remaining training on edge clouds. Specifically, it mathematically describes and evaluates methods to optimize data transfer between GPU memory segments and the overlapping of computational tasks to reduce overall training time and improve GPU utilization. Through various experimental scenarios, we confirmed that asynchronous data transfer and task overlap significantly reduce training time, enhance GPU utilization, and improve model accuracy. In scenarios where all optimization techniques were applied, training time was reduced by 47%, GPU utilization improved to 91.2%, and model accuracy increased to 89.5%. These results demonstrate that asynchronous data transfer and task overlap effectively reduce GPU idle time and alleviate bottlenecks. This study is expected to contribute to the performance optimization of Federated Learning systems in the future.

• Proceedings of the Korea Committee for Ocean Resources and Engineering Conference
• /
• 2004.05a
• /
• pp.24-29
• /
• 2004

A small, highly-trained crew working on the ship's automation has contributed to the improvement of operation efficiency and the labor environment on board ship. However, at the same time, having a small crew adds more responsibility to the ship's officers to safely operate and manage the ship. Recently, development on the system to concentrate important information being scattered at the various pieces of navigational equipment has been actively studied, using information and computer technology. The purpose of this study is to set up and implement an integrated marine data networking and communication system on the training-research ship. Information relating to navigation, engine and office automation were investigated and analyzed, and implementation methods associated with navigation, engine and the management information system were designed and presented. In addition, the networking system and navigational signal interface unit for the integrated communication system, and the data communication method between the ship and land are also discussed.

• Proceedings of the KIEE Conference
• /
• 2003.07a
• /
• pp.181-183
• /
• 2003

This paper presents an operator training simulator for power system restoration against massive black-out. The system is designed especially focused on the generality and convenient setting up for initial condition of simulation. The former is accomplished by using on line calculation methodology, and PSS/E data is used to define the initial situation. The proposed simulator consists of three major components - the Power flow(PF) module, data conversion(CONV) module and CUI subsystem. PF module calculates power flow, and then checks overvoltage of buses and overflow of lines. CONV module composes an Y-Bus array and a data base at each restoration action. The initial Y-Bus array is constructed from PSS/E data. The user friendly GUI module is developed including graphic editor and built-in operation manual. As a result, the maximum processing time for one step operation is 15 seconds, which is adequate for training purpose. Comparison with PSS/E simulation proves the accuracy and reliability of the training system.

• Journal of Ocean Engineering and Technology
• /
• v.18 no.6 s.61
• /
• pp.44-50
• /
• 2004

A small, highly-trained crew working on the ship's automation has contributed to the improvement of operations and the labor environment on board ship. However, at the same time, having a small crew adds more responsibility to the ship's officers to safely operate and manage the ship. With the use of information and computer technology, efforts are being made towards the development of a system that will concentrate important information from the various pieces of navigational equipment. The purpose of this study is to set up and implement an integrated marine data networking and communication system on the training-research ship. Information relating to navigation, engine and office automation are investigated and analyzed, and implementation methods associated with navigation, engine and management information system were designed and presented. In addition, the networking system of the navigational signal interface unit for the integrated communication system, and the data communication method between the ship and land are also discussed.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.28 no.5
• /
• pp.493-504
• /
• 2010

This study proposes a classification method based on an automated training extraction procedure that may be used with very high resolution (VHR) images of non-accessible areas. The proposed method overcomes the problem of scale difference between VHR images and geographic information system (GIS) data through filtering and use of a Landsat image. In order to automate maximum likelihood classification (MLC), GIS data were used as an input to the MLC of a Landsat image, and a binary edge and a normalized difference vegetation index (NDVI) were used to increase the purity of the training samples. We identified the thresholds of an NDVI and binary edge appropriate to obtain pure samples of each class. The proposed method was then applied to QuickBird and SPOT-5 images. In order to validate the method, visual interpretation and quantitative assessment of the results were compared with products of a manual method. The results showed that the proposed method could classify VHR images and efficiently update GIS data.