• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.32 no.5
• /
• pp.975-986
• /
• 2022

Deep learning is attracting great attention, showing excellent performance in image processing, but is vulnerable to adversarial attacks that cause the model to misclassify through perturbation on input data. Adversarial examples generated by adversarial attacks are minimally perturbated where it is difficult to identify, so visual features of the images are not generally changed. Unlikely deep learning models, people are not fooled by adversarial examples, because they classify the images based on such visual features of images. This paper proposes adversarial attack detection method using Symbolic Representation, which is a visual and symbolic features such as color, shape of the image. We detect a adversarial examples by comparing the converted Symbolic Representation from the classification results for the input image and Symbolic Representation extracted from the input images. As a result of measuring performance on adversarial examples by various attack method, detection rates differed depending on attack targets and methods, but was up to 99.02% for specific target attack.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.32 no.2
• /
• pp.405-416
• /
• 2022

Current AI technology is improving the quality of life by using machine learning based on data. When using machine learning, transmitting distributed data and collecting it in one place goes through a de-identification process because there is a risk of privacy infringement. De-identification data causes information damage and omission, which degrades the performance of the machine learning process and complicates the preprocessing process. Accordingly, Google announced joint learning in 2016, a method of de-identifying data and learning without the process of collecting data into one server. This paper analyzed the effectiveness by comparing the difference between the learning performance of data that went through the de-identification process of K anonymity and differential privacy reproduction data using actual financial data. As a result of the experiment, the accuracy of original data learning was 79% for k=2, 76% for k=5, 52% for k=7, 50% for 𝜖=1, and 82% for 𝜖=0.1, and 86% for Federated learning.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.32 no.5
• /
• pp.817-826
• /
• 2022

AI technology is being successfully introduced in many fields, and models deployed as a service are deployed with black box environment that does not expose the model's information to protect intellectual property rights and data. In a black box environment, attackers try to steal data or parameters used during training by using model output. This paper proposes a method of inferring the type of model to directly find out the composition of layer of the target model, based on the fact that there is no attack to infer the information about the type of model from the deep learning model. With ResNet, VGGNet, AlexNet, and simple convolutional neural network models trained with MNIST datasets, we show that the types of models can be inferred using the output values in the gray box and black box environments of the each model. In addition, we inferred the type of model with approximately 83% accuracy in the black box environment if we train the big and small relationship feature that proposed in this paper together, the results show that the model type can be infrerred even in situations where only partial information is given to attackers, not raw probability vectors.

• Journal of Appropriate Technology
• /
• v.5 no.2
• /
• pp.118-125
• /
• 2019

Climate change is one part of 17 Sustainable Development Goals (SDGs). According to the Fifth Assessment Report by the Inter- governmental Panel on Climate Change(IPCC) published in 2014, global warming is caused by greenhouse gas (GHG) emissions. The most important GHG is carbon dioxide (CO₂), which is released by the burning of fossil fuels and, to a lesser extent, by land use practices, followed by nitrous oxide and methane. IPCC predicts that global temperatures will rise 3.7℃ and sea level will rise 0.63 m by 2099 in the case of no strong restraint. According to the report, we can expect a massive species extinctions, changes in storm and drought cycles, altered ocean circulation, and redistribution of vegetation by global warming. However, climate changes, especially global warming, are the largest potential threat to human health and the source of a number of diseases globally. If climate changes are continued uncontrolled, human health will be adversely affected by the accelerating climate change and the natural disaster induced by climate change. It means we will face more serious conditions of injury, disease, and death related to natural disasters such as flood, drought, heat waves, malnutrition, more allergy, air pollution and climate change related infections related to morbidity and mortality. This review emphasizes on the relationship between global climate changes and human health and provides some suggestions for improvement.

• Horticultural Science & Technology
• /
• v.31 no.3
• /
• pp.371-379
• /
• 2013

Anthracnose disease caused by Colletotrichum orbiculare, induces severe damage to cucurbits worldwide. Resistance of 112 commercial cultivars of cucurbits to C. orbiculare was evaluated. Seedlings of each cultivar at 2- to 3-leaf stage were inoculated with C. orbiculare KACC 40809 by spraying spore suspension of the fungus at a concentration of $4.0{\times}10^5$ spores/mL. Among the 36 cultivars of cucumber, 'Asiastrike', 'Tongilbaedadagi', 'Daeseon', 'Cheongrokmatjjang', 'Nokyacheongcheong', and 'Asianogak' were moderately resistant and the others were susceptible. All the tested cultivars of melon (33) and watermelon (4) showed highly susceptible response to C. orbiculare. On the other hand, the squash cultivars (17) represented less susceptibility to the fungus than the other cucurbits. Of the squash cultivars, 'Gammirak' and 'Teotbat' were resistant and 12 cultivars were moderately resistant. Among the rootstocks for cucurbits, ten cultivars including 'JjeuyakaEX', 'Nunbusyeo', 'Union', 'RS111', 'Ganggeuntoza', 'Hwangjaetoza', 'NO.8', 'Shintoza', 'Bulpaetoza', and 'Newtype' showed high resistance to the anthracnose pathogen. From the results, the resistant cultivars could be used as sources of resistance to cucurbits anthracnose (C. orbiculare) in the future breeding programs.

• KIPS Transactions on Computer and Communication Systems
• /
• v.6 no.10
• /
• pp.421-428
• /
• 2017

The need for various approaches to non-face-to-face identification technology for registering and authenticating users online is being required because of the growth of online financial services and the rapid development of financial technology. In general, non-face-to-face approaches can be exposed to a greater number of threats than face-to-face approaches. Therefore, identification policies and technologies to verify users by using various factors and channels are being studied in order to complement the risks and to be more reliable non-face-to-face identification methods. One of these new approaches is to collect and verify a large number of personal information of user. Therefore, we propose a big-data based non-face-to-face Identity Proofing method that verifies identity on online based on various and large amount of information of user. The proposed method also provides an identification information management scheme that collects and verifies only the user information required for the identity verification level required by the service. In addition, we propose an identity information sharing model that can provide the information to other service providers so that user can reuse verified identity information. Finally, we prove by implementing a system that verifies and manages only the identity assurance level required by the service through the enhanced user verification in the non-face-to-face identity proofing process.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.27 no.2
• /
• pp.383-395
• /
• 2017

Brainwave-based user authentication technology has advantages such as changeability, shoulder-surfing resistance, and etc. comparing with conventional biometric authentications, fingerprint recognition for instance which are widely used for smart phone and finance user authentication. Despite these advantages, brainwave-based authentication technology has not been used in practice because of the price for EEG (electroencephalography) collecting devices and inconvenience to use those devices. However, according to the development of simple and convenient EEG collecting devices which are portable and communicative by the recent advances in hardware technology, relevant researches have been actively performed. However, according to the experiment based on EEG samples collected by using a single-channel EEG measurement device which is the most simplified one, the authentication accuracy decreases as the number of channels to measure and collect EEG decreases. Therefore, in this paper, we analyze technical problems that need to be solved for practical use of brainwave-based use authentication and propose an incremental elimination method of collected EEG samples for each user to consist a set of EEG samples which are effective to authentication users.

• Journal of the Korean Chemical Society
• /
• v.63 no.2
• /
• pp.102-110
• /
• 2019

The purpose of this study was to explore the progression levels of science metamodeling knowledge through using questionnaires for 97 students of the gifted in G science academy. As a result of the Rasch model analysis, it was confirmed that the progression levels of the scientific metamodeling knowledge is suitable for the person reliability of 0.71 and the item reliability of 0.96. The progression levels of students' science metamodeling knowledge were classified into 4 stages. First and second levels were considered model to be objective and the third and fourth stages were perceived as subjective. The first level is to view the model as a visual representation of a phenomenon as it is, and the second level is to think that the model corresponds to objective knowledge or theory and is a tool for explanation. The Third level looks at the model as a scientist's exploration tool and fourth level is to think that the model is provisional one and multiple models can coexist in one phenomenon. The progression levels of science metamodeling knowledge of science high school students derived from this study is expected to be used as a reference when constructing a curriculum for science modeling and modeling for gifted students.

• Journal of The Korean Association For Science Education
• /
• v.39 no.3
• /
• pp.457-464
• /
• 2019

This study aims to explore meta-modeling knowledge of gifted students through the modeling. To do this, five gifted students were asked to do modeling related to candle burning, and all the processes of modeling were observed and then individual interviews were conducted. As a result of the study, two students were classified as first level and three students were classified as second level. The students of the first level did not have any model generation or model-based prediction activities, and observation was the most meaningful activity. On the other hand, the students of the second level performed all four modeling processes. However, the generation of the model and the prediction using the model were relatively strong. The data they gained from the experiments was perceived as just confirming the absolute model. No student was found in Level 3 or Level 4. The results of this study show that gifted students remain at the progression level of recognizing the model as an objective reality, and in order to cultivate a true scientist, it is necessary to educate the gifted students to recognize the subjectivity of the model.

• Journal of Appropriate Technology
• /
• v.6 no.1
• /
• pp.37-44
• /
• 2020

Medical waste is any kind of waste that contains infectious material and recommended not to be transferred for infection control. As a means of disposal, incineration has better points than dumping or landfill in the quantity reduction, odorless and nonhazardous. However, open burning and incineration of health care wastes under bad circumstances, can result in the emission of environmental pollutants to air. A burial of biological waste brings pollution of soil and water. Most of sub divisional hospitals in Fiji transfer their medical wastes to divisional hospitals for incineration. In 2011, 62,518 kg of medical waste was incinerated in the three divisional hospitals. However, some medical wastes are considered as general waste and burnt or sent to landfill site, some are buried on site in some sub-divisional hospitals. In this regards, urgent education is necessary for awareness promotion to relevant personnel in medical waste treatment. On site incineration using small scale incinerator is more recommended than transportation of medical wastes treatment in Fiji. Moreover, remotely controllable and fixable small scale of incinerator is more desirable in sub-divisional hospitals. It is recommended that Fiji government to set up a legal framework for medical waste management (MWM), to develop specific guidelines for MWM, to set up a training system for MWM to ensure that all relevant personnel are trained, to develop a monitoring and supervision system for MWM, to clarify the future financing of MWM activities, and to improve the MWM infrastructure.