• Proceedings of the Korea Information Processing Society Conference
• /
• 2024.05a
• /
• pp.711-714
• /
• 2024

In the dynamic landscape of modern machine learning, Federated Learning (FL) has emerged as a compelling paradigm designed to enhance privacy by enabling participants to collaboratively train models without sharing their private data. Specifically, Distillation-based Federated Learning, like Federated Learning with Model Distillation (FedMD), Federated Gradient Encryption and Model Sharing (FedGEMS), and Differentially Secure Federated Learning (DS-FL), has arisen as a novel approach aimed at addressing Non-IID data challenges by leveraging Federated Learning. These methods refine the standard FL framework by distilling insights from public dataset predictions, securing data transmissions through gradient encryption, and applying differential privacy to mask individual contributions. Despite these innovations, our survey identifies persistent vulnerabilities, particularly concerning the susceptibility to logit inversion attacks where malicious actors could reconstruct private data from shared public predictions. This exploration reveals that even advanced Distillation-based Federated Learning systems harbor significant privacy risks, challenging the prevailing assumptions about their security and underscoring the need for continued advancements in secure Federated Learning methodologies.

• International Journal of Computer Science & Network Security
• /
• v.24 no.1
• /
• pp.215-225
• /
• 2024

Predictive maintenance has been considered fundamental in the industrial applications in the last few years. It contributes to improve reliability, availability, and maintainability of the systems and to avoid breakdowns. These breakdowns could potentially lead to system shutdowns and to decrease the production efficiency of the manufacturing plants. The present article aims to study how predictive maintenance could be planed into the production scheduling, through a systematic review of literature. . The review includes the research articles published in international journals indexed in the Scopus database. 165 research articles were included in the search using #predictive maintenance# AND #production scheduling#. Press articles, conference and non-English papers are not considered in this study. After careful evaluation of each study for its purpose and scope, 50 research articles are selected for this review by following the 2020 Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA) statement. A benchmarking of predictive maintenance methods was used to understand the parameters that contributed to improve the production scheduling. The results of the comparative analysis highlight that artificial intelligence is a promising tool to anticipate breakdowns. An additional impression of this study is that each equipment has its own parameters that have to be collected, monitored and analyzed.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.18 no.5
• /
• pp.558-565
• /
• 2015

The current methods, such as Binomial Probability Distribution or T-test, to calculate the number of samples of Precise Guided Missiles(PGMs) for test and evaluation are statistically problematic to make reasonable decisions thus use the budget efficiently. Also, the method to verify an accuracy rate of the PGMs using these old methods seems to be lack of objectiveness. This paper presents an effective, efficient solution adapted from the clinical medicine testing method of healthcare industry research, which may overcome our problems. This method considers and addresses both manufacturers' and consumers' risk simultaneously. By taking into account the weapon system project characteristics which take a relatively longer time and require higher-level budget, this approach also suggests a reasonable and feasible method to determine the amount of samples of the PGMs and making a criterion to verify an accuracy rate of the missiles that are required to test. Thus, this method is expected to contribute to acquisition and operations of high-reliability PGMs by MND and its field-users.

• Science of Emotion and Sensibility
• /
• v.5 no.2
• /
• pp.63-72
• /
• 2002

In order to examine universal design, 1 have developed two analytical methodologies based on 3P(point) task analysis: structured task analysis and task matrix analysis. I also extracted nine universal design items, namely (1) adjustment, (2) redundancy, (3) specification and function transparency, (4) feedback and (5) error tolerance, (6) effective acquisition of information, (7) ease of understanding and judgment, (8) comfortable operation, and (9) continuity of information and operation. Structured task analysis is used to uncover problems in each of the tasks constituting a job for each functionally challenged condition of users, and solutions to the extracted problems are examined in terms of the above-mentioned nine universal design items. Task matrix analysis calls for the production of a table for each task in a job. In each table, nine items form the columns, and the horizontal rows list all disability types. Then, solutions are formulated for each cell formed by the intersecting columns and rows. Using these two analysis methods, 1 have conducted a verification experiment for the universal design of a public bus. The results of the research have enabled me to propose various ,solutions from a system-based perspective, instead of coming up with the superficial and isolated solutions which are normally produced when conventional analytical methods are used.

• Proceedings of the Korean Society for Emotion and Sensibility Conference
• /
• 2002.05a
• /
• pp.144-151
• /
• 2002

In order to examine universal desist I have developed two analytical methodologies based on 3P(point) task analysis: structured task analysis and task matrix analysis. I also extracted me universal design items, namely (1) adjustment (2) redundancy, (3) specification and function transparency, (4) feedback and (5) error tolerance, (6) effective acquisition of information, (7) ease of understanding and judgment (8) comfortable operation, and (9) continuity of information and operation. Structured task analysis is used to uncover problems in each of the tasks constituting a job for each functionally challenged condition of users, and solutions to the extracted problems are examined in terms of the above-mentioned nine universal design items. Task matrix analysis calls for the production of a table for each task in a job. In each table, nine items from the columns, and the horizontal rows list all disability types. Then, solutions are formulated for each cell formed by the intersecting columns and rows. Using these two analysis methods, T have conducted a verification experiment for the universal design of a public bus. The results of the research have enabled me to propose various solutions from a system-based perspective, instead of coming up with the superficial and isolated solutions which are normally produced when conventional analytical methods are used.

• International Journal of Fluid Machinery and Systems
• /
• v.8 no.3
• /
• pp.155-168
• /
• 2015

No-load speed is an important performance factor for the safe operation of hydropower systems. In turbine design, the manufacturers must conduct several model tests to calculate the accurate value of no-load speed for the complete range of operating conditions, which are expensive and time-consuming. The present study presents steady and unsteady methods for calculating no-load speed of a Francis turbine. The steady simulations are implemented using a commercial flow solver and an iterative algorithm that relies on a smooth relation between turbine torque and speed factor. The unsteady method uses unsteady RANS simulations that have been integrated with a user subroutine to compute and return the value of runner speed, time step and friction torque. The main goal of this research is to evaluate and compare the two methods by calculating turbine dynamic parameters for three test cases consisting of high and medium head Francis turbines. Overall, the numerical results agreed well with experimental data. The unsteady method provided more accurate results in the opening angle range from 20 to 26 degrees. Nevertheless, the steady results showed more consistency than unsteady results for the three different test cases at different operating conditions.

• Spatial Information Research
• /
• v.7 no.1
• /
• pp.49-61
• /
• 1999

The purpose of this study is to find out the methodologies, using GIS techniques, to appraise the parcel prices and to estimate the compensation expenditures according to road expansions. This study consists of 2 parts. The first pare is to find out the ways for appraisal of patrol prices in certain area by the publicly announced land prices. The second part is to estimate the differences of compensation expenditures according to road expansion methods. The conclusions are as follows. First, to appraise the target parcel prices, a comparison to the publicly announced land prices should be used. For this, 19 items of land characteristics of existing land information should be analyzed and compared using GIS techniques. Second, the compensation expenditure could be easily estimated using GIS techniques too. In this process, if necessary, we have to apply the minimum parcel size in Building Codes.

• Korean Journal of Oriental Medicine
• /
• v.13 no.2 s.20
• /
• pp.127-133
• /
• 2007

Objectives : This study were aimed to estimate the needs for clinical research educational program development focused on oriental medicine. Methods : Department Medical Research of Korea Institute of Oriental Medicine surveyed 204 certified subjects in total by using web-based questionnaires through e-mail from Jan. 17th to Jan. 31th, 2007. Reply from 62 on-line correspondents were collected and statistically analyzed. Results : The number of clinical trial involving oriental medicine continues to increase. According to the survey, many Issues were raised as problems such as difficulty of recruitment, lack of fund and lack of stepped program. Emphasized issues were clinical research methodologies, development of protocols and case report form (CRF), and Regulations including Institutional Review Boards (IRBs) in these three consecutive education training program. Conclusion : The results of this study may contribute to the development of an educational program for oriental medicine, a program that should be taken into consideration for developing practical items, such as, problem-based learnings which reflects participants' actual needs in their works. Also this report be used for future strategy plans and execution of training program for oriental medical clinical trial.

• Parasites, Hosts and Diseases
• /
• v.61 no.4
• /
• pp.418-427
• /
• 2023

Toxoplasma gondii infections are primarily diagnosed by serological assays, whereas molecular and fluorescence-based techniques are garnering attention for their high sensitivity in detecting these infections. Nevertheless, each detection method has its limitations. The toxoplasmosis detection capabilities of most of the currently available methods have not been evaluated under identical experimental conditions. This study aimed to assess the diagnostic potential of enzyme-linked immunosorbent assay (ELISA), real-time polymerase chain reaction (RT-PCR), immunohistochemistry (IHC), and immunofluorescence (IF) in BALB/c mice experimentally infected with various doses of T. gondii ME49. The detection of toxoplasmosis from sera and brain tissues was markedly enhanced in mice subjected to high infection doses (200 and 300 cysts) compared to those subjected to lower doses (10 and 50 cysts) for all the detection methods. Additionally, increased B1 gene expression levels and cyst sizes were observed in the brain tissues of the mice. Importantly, IHC, IF, and ELISA, but not RT-PCR, successfully detected T. gondii infections at the lowest infection dose (10 cysts) in the brain. These findings may prove beneficial while designing experimental methodologies for detecting T. gondii infections in mice.

• Korean Chemical Engineering Research
• /
• v.59 no.3
• /
• pp.316-325
• /
• 2021

Human-made plastics takes more than 500 years to break down and have been introduced into the ecosystem, threatening terrestrial and aquatic organisms. By 2025, there will be more than 250 million tons of plastics in the ocean. Although studies regarding microplastics have been exponentially increasing since 2015, international standards for defining the size classification for microplastics, as well as methods for qualitative/quantitative analysis have not been concluded yet. Thus, in this study, the latest trends in analytical methodologies have been reviewed. Further, the study suggests directions for future research approaches can be taken to analyze aquatic microplastics, which could be as useful information for establishing effective microplastic management policy via standardization in microplastic analysis.