• Proceedings of the Korea Information Processing Society Conference
• /
• 2024.05a
• /
• pp.69-72
• /
• 2024

모델 역추론 공격은 공격 대상 네트워크를 훈련하기 위해 사용되는 훈련 데이터셋 중 개인 데이터셋을 공개 데이터셋을 사용하여 개인 훈련 데이터셋을 복원하는 것이다. 모델 역추론 방법 중 적대적 생성 신경망을 사용하여 모델 역추론 공격을 하는 과거의 논문들은 딥러닝 모델 전체의 역추론에만 초점을 맞추기 때문에, 이를 통해 얻은 원본 이미지의 개인 데이터 정보는 제한적이다. 따라서, 본 연구는 대상 모델의 중간 출력을 사용하여 개인 데이터에 대한 더 품질 높은 정보를 얻는데 초점을 맞춘다. 본 논문에서는 적대적 생성 신경망 모델이 원본 이미지를 생성하기 위해 사용되는 계층별 역추론 공격 방법을 소개한다. MNIST 데이터셋으로 훈련된 적대적 생성 신경망 모델을 사용하여, 원본 이미지가 대상 모델의 계층을 통과하면서 얻은 중간 계층의 출력 데이터를 기반으로 원본 이미지를 재구성하고자 한다. GMI 의 공격 방식을 참고하여 공격 모델의 손실 함수를 구성한다. 손실 함수는 사전 손실 및 정체성 손실항을 포함하며, 역전파를 통해서 원본 이미지와 가장 유사하게 복원할 수 있는 표현 벡터 Z 를 찾는다. 원본 이미지와 공격 이미지 사이의 유사성을 분류 라벨의 정확도, SSIM, PSNR 값이라는 세 가지 지표를 사용하여 평가한다. 공격이 이루어지는 계층에서 복원한 이미지와 원본 이미지를 세 가지 지표를 가지고 평가한다. 실험 결과, 공격 이미지가 원본 이미지의 대상 분류 라벨을 정확하게 가지며 원본 이미지의 필체를 유사하게 복원하였음을 보여준다. 평가 지표 또한 원본 이미지와 유사하다는 것을 나타낸다.

• Journal of the Korean Society of Marine Environment & Safety
• /
• v.25 no.4
• /
• pp.423-432
• /
• 2019

Along with an increase in marine activities every year, the frequency of marine accidents is on the rise. Accordingly, various research activities and policies for marine safety are being implemented. Despite these efforts, the number of accidents are increasing every year, bringing their effectiveness into question. Preliminary studies relying on annual statistical reports provide precautionary measures for items that stand out significantly, through the comparison of statistical provision items. Since the 2000s, large-scale marine accidents have repeatedly occurred, and case studies have examined the "accident response." Likewise, annual statistics or accident cases are used as core data in policy formulation for domestic maritime safety. However, they are just a summary of post-accident results. In this study, limitations of current marine research and policy are evaluated through a literature review of case studies and analyses of marine accidents. In addition, the ontology of the marine accident information classification system will be revised to improve the current limited usage of the information through an attribute analysis of boating accident status reports and text mining. These aspects consist of the reporter, the report method, the rescue organization, corrective measures, vulnerability of response, payloads, cause of oil spill, damage pattern, and the result of an accident response. These can be used consistently in the future as classified standard terms to collect and utilize information more efficiently. Moreover, the research proposes a data collection and quality assurance method for the practical use of ontology. A clear understanding of the problems presently faced in marine safety will allow "suf icient quality information" to be leveraged for the purpose of conducting various researches and realizing effective policies.

• Korean Association of Arts Management
• /
• no.54
• /
• pp.95-132
• /
• 2020

Many Korean traditional music such as Jongmyo Jeryeak and Pansori are listed as UNESCO World Intangible Cultural Heritage and are recognized for their value. However, compared to other genres, Korean traditional music shows various problems such as low proportion of paying audiences and bias of audience age. Previous studies in the performing arts point out that in order to solve these problems, it is necessary to analyze consumers. Existing previous studies on the Korean traditional music performance remained mainly to grasp the influence of performance quality of service on viewing satisfaction at the one-dimensional level. However, in order to meet consumer expectations, not only the satisfaction factor but also the reasons why performing art consumers do not consume performances. In addition, it is important to identify the factors that can cause service failure before the performance, to prepare countermeasures and to manage them continuously. In this study, the Kano-Timko model was used to classify the quality factors of Korean traditional music performances, and to identify the priority points of service failures of Korean traditional music performances through the potential customer satisfaction index. Through the results of this study, Korean traditional performing organizations and organizations will be able to prepare proactive countermeasures for service failures occurring in Korean traditional music performances and establish effective operational strategies to derive customer satisfaction. In addition, it is important to apply the concept of service failure in order to classify the performance quality of service that was identified only in a one-dimensional manner based on Kano model.

• The Journal of Society for e-Business Studies
• /
• v.14 no.4
• /
• pp.1-17
• /
• 2009

E-Catalogs which describe products or services are one of the most important data for the electronic commerce. E-Catalogs are created, updated, and removed in order to keep up-to-date information in e-Catalog database. However, when the number of catalogs increases, information integrity is violated by the several reasons like catalog duplication and abnormal classification. Catalog search, duplication checking, and automatic classification are important functions to utilize e-Catalogs and keep the integrity of e-Catalog database. To implement these functions, probabilistic models that use statistics of index words extracted from e-Catalogs had been suggested and the feasibility of the methods had been shown in several papers. However, even though these functions are used together in the e-Catalog management system, there has not been enough consideration about how to share common data used for each function and how to effectively manage statistics of index words. In this paper, we suggest a method to implement these three functions by using simple SQL supported by relational database management system. In addition, we use materialized views to reduce the load for implementing an application that manages statistics of index words. This brings the efficiency of managing statistics of index words by putting database management systems optimize statistics updating. We showed that our method is feasible to implement three functions and effective to manage statistics of index words with empirical evaluation.

• Food Science of Animal Resources
• /
• v.23 no.4
• /
• pp.327-332
• /
• 2003

This study was conducted to investigate the characteristics of strains which were isolated from market milk treated with ESL(extended shelf life) and conventional system, and to compare the microbiological quality of ESL milk with conventional milk. In order to characterize the isolated strains, purification, Gram staining, spore staining, catalase, oxidase, motility test, and identification by means of automatic identificator were performed. The results obtained are as follows: total 364 selected strains were analyzed in this study. Depending upon the isolated source, the number of strains from conventional milk was found to be Higher than ESL-milk. By means of grouping of total strains, Bacillus ssp. and Staphylococcus ssp. showed to be predominant. But most of strains were distributed with various groups except Lactobacillus ssp. When the isolates were compared with milk process methods, Enterococcus ssp. was detected much on market milk treated with LTLT pasteurization. Also, Pseudomonas ssp. was detected much on conventional milk treated with UHT pasteurization. By comparison with genus groups depending upon storage temperature of market milk, the higher milk storage temperature increased, the most frequency detected Bacillus ssp. increased. Also, Pseudomonas ssp. was detected most frequently at 10$^{\circ}C$ storage condition. Generally this genus derived from post-contamination during milk processing and related to the quality of market milk during chilled system. In conclusion, it was shown that ESL system reduced post-contamination during milk process, following the improvement of product quality and life cycle during the distribution of market milk.

• The Journal of the Convergence on Culture Technology
• /
• v.8 no.2
• /
• pp.385-391
• /
• 2022

Quality evaluation of beef carcasses is an important issue in the livestock industry. Recently, through the AI monitor system based on artificial intelligence, the quality manager can receive help in making accurate decisions based on the analysis of beef carcass images or result information. This artificial intelligence dataset is an important factor in judging performance. Existing datasets may have different surface orientation or resolution. In this paper, we proposed a two-stage classification model that can efficiently manage the grades of beef carcass image using deep learning. And to overcome the problem of the various conditions of the image, a new dataset of 1,300 images was constructed. The recognition rate of deep network for 5-grade classification using the new dataset was 72.5%. Two-stage evaluation is a method to increase reliability by taking advantage of the large difference between grades 1++, 1+, and grades 1 and 2 and 3. With two experiments using the proposed two stage model, the recognition rates of 73.7% and 77.2% were obtained. As this, The proposed method will be an efficient method if we have a dataset with 100% recognition rate in the first stage.

• Journal of Nuclear Fuel Cycle and Waste Technology(JNFCWT)
• /
• v.4 no.1
• /
• pp.65-75
• /
• 2006

Transparency on the Total System Performance Assessment (TSPA) is the key issue to enhance the public acceptance for a radioactive repository. To approve it, all performances on TSPA through Quality Assurance is necessary. The integrated Cyber R&D Platform is developed by KAERI using the T2R3 principles applicable for five major steps : planning, research work, documentation, and internal & external audits in R&D's. The proposed system is implemented in the web-based system so that all participants in TSPA are able to access the system. It is composed of three sub-systems; FEAS (FEp to Assessment through Scenario development) showing systematic approach from the FEPs to Assessment methods flow chart, PAID (Performance Assessment Input Databases) being designed to easily search and review field data for TSPA and QA system containing the administrative system for QA on five key steps in R&D's in addition to approval and disapproval processes, corrective actions, and permanent record keeping. All information being recorded in QA system through T2R3 principles is integrated into Cyber R&D Platform so that every data in the system can be checked whenever necessary. Throughout the next phase R&D, Cyber R&D Platform will be connected with the assessment tool for TSPA so that it will be expected to search the whole information in one unified system.

• Culinary science and hospitality research
• /
• v.21 no.4
• /
• pp.294-308
• /
• 2015

The purpose of this study was to look into the service quality of websites and the attributes of review information thereon as perceived by visitors thereto and to analyze the effect that the service quality of festival websites and informative usefulness of testimonials would have on the satisfaction with use, along with the effect that the satisfaction with use would have on behavioral intention. Based on the results of this study, we intended to present measures that could help strengthen the intention to spread positive word-of-mouth for festivals and to visit a festival through satisfaction with a festival's website. The festival selected to achieve the objectives of this study was the Yeongdeok Snow Crab Festival, which is one of the most renowned local food festivals in the Gyeongbuk region, having been held for 17 consecutive years, and has been selected as the top festival of Gyeongsangbuk-do for 4 consecutive years. We surveyed visitors to Yeongdeok Snow Crab Festival who had searched for information on the Yeongdeok Snow Crab Festival prior to its opening, as well as those who had participated in Yeongdeok Snow Crab Festival before. The results of this study showed that review information on a food festival website, service quality design, adequacy of substance, responsiveness to customers, and system capability were found to have a positive effect on behavioral intention through satisfaction with use. The findings suggest that it is crucial to manage the service quality of websites in order to stimulate interest in, and induce visits to, food festivals.

• Food Science and Preservation
• /
• v.24 no.6
• /
• pp.771-777
• /
• 2017

Although Doenjang is a representative traditional fermented food in Korea, made mainly from soybeans, it has been classified into a food group identified as having high sodium in the National Health and Nutrition Examination. It is necessary to develop a low sodium Doenjang to prevent the excessive sodium intake which may cause various diseases. However, the development of Doenjang with low sodium, without significant changes on quality, is an ongoing challenge. Therefore, the experiment was designed to reduce the salt content of the soaking water to 12.5-45%. The pH, saltiness. moisture, sodium, color, amino-type nitrogen, free amino acids and constituent amino acids composition of Doenjang were investigated to determine the effect of this salt reduction on the sensory quality of Doenjang. The reduction of sodium did not affect the pH, moisture and saltiness, and this changed maintained the same range as the control. The sodium content was reduced proportionally, and the amino acid nitrogen level was 500 mg%. Therefore, this study considered that a reduction of 25% of salt in the soaking water does not affect the quality of the Doenjang while reducing the sodium content of the final Doenjang.

• Journal of agriculture & life science
• /
• v.45 no.6
• /
• pp.291-301
• /
• 2011

This study is based on the diagnosis process on agricultural production cost survey, which is one of the most representative public statistics of agricultural sector in Korea. The diagnosis were performed by employing the quantitative and qualitative approaches with the official guideline provided by the Ministry of Statistics. The various results were produced from the aspects of exactness, relevance, timeliness, accessibility, and comparability, which were based on the qualitative approaches including focus group interview (FGI) with professional users. The results from statistical portfolio analysis implies that the first priority to improve the statistics is given to 'satisfaction level corresponding to user cost' and 'offering sufficient explanation'. The comprehensive results, discussed with the administrative office, are summarized into the following two major points. First, classification of the detailed items should be improved to avoid users' confusion. However, the re-classification need to be professionally reviewed with the relevant administrative office such as Rural Development Administration, producing similar statistics. Second, the KOSIS system is found to be constantly upgraded to improve the users' convenience.