• Journal of Korea Multimedia Society
• /
• v.25 no.6
• /
• pp.878-885
• /
• 2022

Research in the field of computer vision based on deep learning is being actively conducted. However, deep learning-based models have vulnerabilities in adversarial attacks that increase the model's misclassification rate by applying adversarial perturbation. In particular, in the case of FGSM, it is recognized as one of the effective attack methods because it is simple, fast and has a considerable attack success rate. Meanwhile, as one of the efforts to visualize deep learning models, Grad-CAM enables visual explanation of convolutional neural networks. In this paper, I propose a method to generate adversarial examples with high attack success rate by applying Grad-CAM to FGSM. The method chooses fixels, which are closely related to labels, by using Grad-CAM and add perturbations to the fixels intensively. The proposed method has a higher success rate than the FGSM model in the same perturbation for both targeted and untargeted examples. In addition, unlike FGSM, it has the advantage that the distribution of noise is not uniform, and when the success rate is increased by repeatedly applying noise, the attack is successful with fewer iterations.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.32 no.4
• /
• pp.617-628
• /
• 2022

Ascon is one of the final round candidates of the NIST lightweight cryptography contest, which has been underway since 2015, and supports hash modes Ascon-Hash and Ascon-Xof. In this paper, we develop a MILP model for collision attack on the Ascon-Hash and search for a differential trail that can be used in a quantum setting through the model. In addition, we present an algorithm that allows an attacker who can use a quantum computer to find a quantum free-start collision attack of 3-round Ascon-Hash using the discovered differential trail. This attack is meaningful in that it is the first to analyze a collision attack on Ascon-Hash in a quantum setting.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.37 no.4B
• /
• pp.288-299
• /
• 2012

To ensure data confidentiality and fine-grained access control in business environment, system model using KP-ABE(Key Policy-Attribute Based Encryption) and PRE(Proxy Re-Encryption) has been proposed recently. However, in previous study, data confidentiality has been effected by decryption right concentrated on cloud server. Also, Yu's work does not consider a access privilege management, so existing work become dangerous to collusion attack between malicious user and cloud server. To resolve this problem, we propose secure system model against collusion attack through dividing data file into header which is sent to privilege manager group and body which is sent to cloud server and prevent modification attack for proxy re-encryption key using d Secret Sharing, We construct protocol model in medical environment.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.29 no.1
• /
• pp.117-125
• /
• 2019

Deep Learning, which is being used in various fields recently, is being threatened with Adversarial Attack. In this paper, we experimentally verify that the classification accuracy is lowered by adversarial samples generated by malicious attackers in image classification models. We used MNIST dataset and measured the detection accuracy by injecting adversarial samples into the Autoencoder classification model and the CNN (Convolution neural network) classification model, which are created using the Tensorflow library and the Pytorch library. Adversarial samples were generated by transforming MNIST test dataset with JSMA(Jacobian-based Saliency Map Attack) and FGSM(Fast Gradient Sign Method). When injected into the classification model, detection accuracy decreased by at least 21.82% up to 39.08%.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.7
• /
• pp.1916-1934
• /
• 2023

The phishing attack is a malicious emerging threat on the internet where the hackers try to access the user credentials such as login information or Internet banking details through pirated websites. Using that information, they get into the original website and try to modify or steal the information. The problem with traditional defense systems like firewalls is that they can only stop certain types of attacks because they rely on a fixed set of principles to do so. As a result, the model needs a client-side defense mechanism that can learn potential attack vectors to detect and prevent not only the known but also unknown types of assault. Feature selection plays a key role in machine learning by selecting only the required features by eliminating the irrelevant ones from the real-time dataset. The proposed model uses Hyperparameter Optimized Artificial Neural Networks (H-ANN) combined with a Hybrid Firefly and Grey Wolf Optimization algorithm (H-FFGWO) to detect and block phishing websites in Internet of Things(IoT) Applications. In this paper, the H-FFGWO is used for the feature selection from phishing datasets ISCX-URL, Open Phish, UCI machine-learning repository, Mendeley website dataset and Phish tank. The results showed that the proposed model had an accuracy of 98.07%, a recall of 98.04%, a precision of 98.43%, and an F1-Score of 98.24%.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.15 no.6
• /
• pp.1919-1932
• /
• 1991

Experimental studies are conducted to investigate the Flow-Induced Vibration mechanism for cantilever plate model with the angle of attack (.alpha.=10.deg., 20.deg., 30.deg.). Research is divided into two parts. First, the flow fields around two dimensional flat plate model are investigated using LDV system. Second, the vortex shedding frequency and response spectra of cantilever plate are obtained experimentally using gap sensor and hot wire anemometer. Finite element method program was used in order to predict the flow field and pressure field around thin flat plate. And some predicted results were compared with the experimental data. The aspect ration of test model is d/t=25 (d; width, t; thickness). From the measurement of the flow field it was found that in the case of small inclined (.alpha.=10.deg., 20.deg.) relatively, the separated boundary layer at sharp leading edge developed smoothly downstream. With increasing the angle of attack of the plate, stagnation region was appeared on the back side of the plate and separated boundary layer was extended downstream. These trends are a good agreement with the computational results. It was found by analysis of response spectra of cantilever plate that the influences of vortex shedding frequency were important at the large of attack (.alpha.=30.deg.), and two peak values appear in entire test model at 24Hz, 150Hz.

• The Journal of the Korea Contents Association
• /
• v.21 no.4
• /
• pp.262-277
• /
• 2021

In order to explore the power process, a hypothetical model which explains the interrelationships among 3 marital power(traditional, egalitarian, personal), 3 influence strategies(reward, coercion, emotional), and 2 negative conflict resolution styles(attack vs. avoidance) was developed. In order to examine the gender differences, male model and female model were developed separately and compared. Using the data collected from 182 males and 196 females, the hypothetical model was tested. For data analysis, SEM was used. As a result, 3 common paths were found: Greater use of emotional influence strategy increased attack as well as avoidance. Greater egalitarian power increased reward. Egalitalian power affected the use of coercion, but the direction was opposed: male's egalitarian power decreased coercion, while female's egalitarian power increased it. Except these, the analyses revealed the substantial differences between male and female. Based on the findings, the ways to reduce attack and avoidance, and theoretical implications were discussed.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.32 no.5
• /
• pp.997-1008
• /
• 2022

Today, AI(Artificial Intelligence) technology is being extensively researched in various fields, including the field of malware detection. To introduce AI systems into roles that protect important decisions and resources, it must be a reliable AI model. AI model that dependent on training dataset should be verified to be robust against new attacks. Rather than generating new malware detection, attackers find malware detection that succeed in attacking by mass-producing strains of previously detected malware detection. Most of the attacks, such as adversarial attacks, that lead to misclassification of AI models, are made by slightly modifying past attacks. Robust models that can be defended against these variants is needed, and the Robustness level of the model cannot be evaluated with accuracy and recall, which are widely used as AI evaluation indicators. In this paper, we experiment a framework to evaluate robustness level by generating an adversarial sample based on one of the adversarial attacks, C&W attack, and to improve robustness level through adversarial training. Through experiments based on malware dataset in this study, the limitations and possibilities of the proposed method in the field of malware detection were confirmed.

• Journal of the Korean Society for Aviation and Aeronautics
• /
• v.24 no.3
• /
• pp.55-61
• /
• 2016

Among the method of aerial fire firefighting, the indirect attack is efficiency way to protect main facilities and it is the aerial fire line construction. According to this study is suggested the fire line construction strategy of indirect attack by helicopter suitable Korea forest fire on theory consideration of indirect attack and experience in practical scene. This study defined that main key points of the fire line construction are accuracy, large quantity, and quickness. Main protection facilities are devided as caution area, warning area, danger area and concern area. Also, it suggested stage-by-stsge from 1 step to 3 step for the aerial fire firefighting correspondence strategy and the fire line construction model. I regard that this study's indirect attack method of the aerial fire firefighting of the fire line construction may be understand about indirect attack tactics and application of indirect attack which is assistance to raise of capability of the aerial fire firefighting with effectiveness and efficiency.

• Wind and Structures
• /
• v.30 no.5
• /
• pp.499-509
• /
• 2020

The aeroelastic stability of a long-span suspension footbridge with a bluff deck (prototype section) was examined through static and dynamic wind tunnel tests using a 1:10 scale sectional model of the main girder, and the corresponding aerodynamic countermeasures were proposed in order to improve the stability. First, dynamic tests of the prototype sectional model in vertical and torsional motions were carried out at three attack angles (α = 3°, 0°, -3°). The results show that the galloping instability of the sectional model occurs at α = 3° and 0°, an observation that has never been made before. Then, the various aerodynamic countermeasures were examined through the dynamic model tests. It was found that the openings set on the vertical web of the prototype section (web-opening section) mitigate the galloping completely for all three attack angles. Finally, static tests of both the prototype and web-opening sectional models were performed to obtain the aerodynamic coefficients, which were further used to investigate the galloping mechanism by applying the Den Hartog criterion. The total damping of the prototype and web-opening models were obtained with consideration of the structural and aerodynamic damping. The total damping of the prototype model was negative for α = 0° to 7°, with the minimum value being -1.07%, suggesting the occurrence of galloping, while that of the web-opening model was positive for all investigated attack angles of α = -12° to 12°.