• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.32 no.1
• /
• pp.89-98
• /
• 2022

As deep learning technology was applied to various fields, research on adversarial attack techniques, a security problem of deep learning models, was actively studied. adversarial attacks have been mainly studied in the field of images. Recently, they have even developed a complete decision-based attack technique that can attack with just the classification results of the model. However, in the case of the audio field, research is relatively slow. In this paper, we applied several decision-based attack techniques to the audio field and improved state-of-the-art attack techniques. State-of-the-art decision-attack techniques have the disadvantage of requiring many queries for gradient approximation. In this paper, we improve query efficiency by proposing a method of reducing the vector search space required for gradient approximation. Experimental results showed that the attack success rate was increased by 50%, and the difference between original audio and adversarial examples was reduced by 75%, proving that our method could generate adversarial examples with smaller noise.

• Journal of Internet Computing and Services
• /
• v.25 no.2
• /
• pp.1-9
• /
• 2024

Although federated learning is designed to be safer than centralized methods in terms of security and privacy, it still has many vulnerabilities. An attacker performing an adversarial attack intentionally manipulates the deep learning model by injecting carefully crafted input data, that is, adversarial examples, into the client's training data to induce misclassification. A common defense strategy against this is so-called adversarial training, which involves preemptively learning the characteristics of adversarial examples into the model. Existing research assumes a scenario where all clients are under adversarial attack, but considering the number of clients in federated learning is very large, this is far from reality. In this paper, we experimentally examine aspects of adversarial training in a scenario where some of the clients are under attack. Through experiments, we found that there is a trade-off relationship in which the classification accuracy for normal samples decreases as the classification accuracy for adversarial examples increases. In order to effectively utilize this trade-off relationship, we present a method to perform adversarial training by adaptively selecting a loss function depending on whether the client is attacked.

• Structural Engineering and Mechanics
• /
• v.30 no.3
• /
• pp.331-350
• /
• 2008

A framework for multi-platform analytical and multi-component hybrid (testing-analysis) simulations is described in this paper and illustrated with several application examples. The framework allows the integration of various analytical platforms and geographically distributed experimental facilities into a comprehensive pseudo-dynamic hybrid simulation. The object-oriented architecture of the framework enables easy inclusion of new analysis platforms or experimental models, and the addition of a multitude of auxiliary components, such as data acquisition and camera control. Four application examples are given, namely; (i) multi-platform analysis of a bridge with soil and structural models, (ii) multiplatform, multi-resolution analysis of a high-rise building, (iii) three-site small scale frame hybrid simulation, and (iv) three-site large scale bridge hybrid simulation. These simulations serve as illustrative examples of collaborative research among geographically distributed researchers employing different analysis platforms and testing equipment. The versatility of the framework, ease of including additional modules and the wide application potential demonstrated in the paper provide a rich research environment for structural and geotechnical engineering.

• Communications of Mathematical Education
• /
• v.21 no.2 s.30
• /
• pp.125-152
• /
• 2007

When we teach mathematics in college, we find a lot of mistakes, fallacies and flaws in the solution of the students. In this paper, we presented a variety of examples of mistakes and fallacies, including wrong proofs, misinterpreted definitions and the mistaken use of theory. The examples, taken from different classes and subjects, are based on our own experience of teaching mathematics. As the previous research argued, such mistakes, fallacies and flaws should be considered as natural phenomena in the students' progress and should be analyzed systematically for the more effective education. By providing a wide-ranging examples of mistakes and fallacies, and detailed analyses of them, we emphasized the significance of the analysis of mistakes and fallacies and proposed that more careful attention should be paid on the collection and development of teaching materials in the area of mistake and fallacy analysis. We hope that this study would be a meaningful contribution to the teaching of mathematics in college.

• Journal of the Korea Fashion and Costume Design Association
• /
• v.12 no.1
• /
• pp.103-116
• /
• 2010

In view of our contemporary fashion, it is found that crossover between clothing and other items leads to creation of a new design and crossover of various cultural codes, ultimately creating a multicultural clothing design or any external outline of emerging costume designs by attempting mutual combination with other fields. Starting from a viewpoint that mutual combination style has significant effects on our contemporary fashion designs, this study intends to characterize external aspects of fashion design that changes through mutual combination style. This study focused upon analyzing costumes released by contemporary fashion designers after 2000, and addressed a variety of mutual combination styles. It gives various examples on mutual combination in fashion, seeking first to look into typical examples of mutual combination styling between fashion and art, between fashion and space and between fashion and technology. Based on those examples, this study classified crossover styles into 6 major categories such as attachment, suspension, modification, fusion, association and embedment. As a result, this study comes to a conclusion that external changes by mutual combination are characterized largely by overlap, simultaneity and deconstruction.

• Journal of the Korean School Mathematics Society
• /
• v.13 no.1
• /
• pp.143-161
• /
• 2010

The purpose of this paper was to analyze the example spaces related with the concepts of triangles in the 2nd grade Korean elementary mathematics textbooks, investigate the meaning of example spaces, and suggest the implications to teaching and learning geometrical concepts in the elementary schools. Mathematics textbooks are usually major materials in teaching and learning mathematics in Korea, and the content of the textbooks affects the quality of teaching and learning in the classroom. The results of the study showed that the Korean elementary textbooks provided examples with limited example spaces concerning the concepts of triangles. The researcher suggested that we should provide rich examples spaces, encourage learner-generated example spaces, and strengthen the connections of examples between in textbooks and in the classroom.

• Proceedings of the Korean Institute of Interior Design Conference
• /
• 2003.05a
• /
• pp.74-77
• /
• 2003

After the Enforcement of Price Deregulation of Apartment, Apartment house get down to originality goods, The Housing Market have reorganized the nucleus by a user, have demanding the development for discriminative unit plan. The purpose of this study is that before and after the Price Decontrol of Apartment take part a variety of unit plan, search for transformation factor and analyze into the tendency of the distinction plan of Housing Goods. Before and after the Price Decontrol of Apartment, Apartment unit have analyzed from 85 $m^2$ till 152 $m^2$ private area; ten corporations of civil construction' unit in Seoul and The national capital region supply apartment, will supply apartment. For selected examples, first, unit plan is normalized from the ratio of front to side wall, bay, a Room' organization and a kind of Room, number, and for examples of unit plan of apartment, the examples were analyzed with respect to change of a Room' organization and the number of a room and the ratio of front wall to side wall for item investigated. Finally, I search out course of transformation tendency of an apartment unit plan after Enforcement of Price Deregulation and analyzed a factor. The results of the study are follows, after Enforcement of Price Deregulation, unit plan of apartment lead to change lay out, to secure each family's privacy, to secure feeling for open hearted, tendency of flexibility.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.31 no.11
• /
• pp.1059-1068
• /
• 2007

Nonlinear response structural optimization is performed using equivalent loads (NROEL). Nonlinear response optimization is extremely cost because many nonlinear analyses are required. In NROEL, the external loads are transformed to the equivalent loads (EL) for linear static analysis and linear response optimization is carried out based on the EL in a cyclic manner until the convergence criteria are satisfied. EL is the load set which generates the same response field of linear analysis as that of nonlinear analysis. The primitive from of theory has been published. In this research, the theory is investigated with large scale example problems. Four examples are solved by using NROEL. Conventional optimization with sensitivity analysis using the finite difference method (FDM) is also applied to the same examples. Moreover, response surface optimization method is applied to the last two examples. The results of the optimizations are compared. In nonlinear response optimization of large scale problems, hundreds (or even thousands) of nonlinear analyses are expected to satisfy the convergence criteria. However, in nonlinear response optimization using equivalent loads, only tens of nonlinear analyses are required. The results are discussed and the usefulness of NROEL is presented.

• Journal of Korean Medical classics
• /
• v.26 no.4
• /
• pp.305-323
• /
• 2013

Through the study of Bohenggyuljangbuyongyakbubyo(輔行訣臟腑用藥法要) focusing on the newly illuminated subject 'The study of Donhuang(敦煌學)' of China, the author analyzed and researched Donhuangzhongyiyakjunseo Bohenggyuljangbuyongyakbubyo(敦煌中醫藥全書 輔行訣臟腑用 藥法要) in order to study and report the origin of prescriptions from Sanghanjabbyungron(傷寒雜病論), which was written by Zhangzhongjing(張仲景). Among around 50 prescriptions included in Bohenggyul, 18 prescriptions similar to Sanghanjabbyungron could be classified in five different manners as follows. 1) There are nine examples whose prescription composition and chief virtue are same but names different. 2) There are five examples whose prescription composition is same but chief virtue and names different. 3) There are two examples whose names are same but prescription composition and chief virtue different. 4) There is one example of prescription as a combination of two or more prescriptions. 5) There is one example whose prescription composition and possible variations of prescription are same but chief virtue different.

• Journal of the Korean Institute of Telematics and Electronics A
• /
• v.31A no.10
• /
• pp.119-131
• /
• 1994

In this paper, we first propose a new FPGA routing method in which seversal netw are routed concurrently by applying the readitional maze routing method. We then introduce CMRF (concurrent maze Router for FPGA) which can be used for the routing of FpGAs of symmetrical array type by applying our new routing method. Given a set of nets, the proposed routing method performas the maze propagation and backtracing independently for each net and determines the routing paths concurrently by competition among nets. In CMRF, using this routing method, q nets are selected from the nets to be routed and they are routed concurrently, where q is the user given parameter determined by considering the computing environment. This process is repeated until either all the nets are routed or the remaining unrouted nets fail to their maze propagations. The routing of these nets are completed using the rip-up and rerouting technique. We apply our routing method to ten randomly generated test examples in order to check its routing performance. The results show taht as we increase the value of q, the routing completion rate increases for all the examples. Note that when q=1, our method is similar to the conventinal maze routing method. We also compare CMRF with the CGE method which has been proposed by Brown et.al. For the five benchmark examples, CMRF complete the routing with less wire segments in each connection block than the wire segments needed in the CGE method of 100% routing.