• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.30 no.5
• /
• pp.871-879
• /
• 2020

Perceptual Ad-Blocking is a new advertising blocking technique that detects online advertising by using an artificial intelligence-based advertising image classification model. A recent study has shown that these Perceptual Ad-Blocking models are vulnerable to adversarial attacks using adversarial examples to add noise to images that cause them to be misclassified. In this paper, we prove that existing perceptual Ad-Blocking technique has a weakness for several adversarial example and that Defense-GAN and MagNet who performed well for MNIST dataset and CIFAR-10 dataset are good to advertising dataset. Through this, using Defense-GAN and MagNet techniques, it presents a robust new advertising image classification model for adversarial attacks. According to the results of experiments using various existing adversarial attack techniques, the techniques proposed in this paper were able to secure the accuracy and performance through the robust image classification techniques, and furthermore, they were able to defend a certain level against white-box attacks by attackers who knew the details of defense techniques.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.22 no.8
• /
• pp.1814-1822
• /
• 1997

In this paper, we present a new digital signature for copyright protection of digital images. The proposed algorithm is designed to be more robust to both the compression (quantization) errors and the illegal signature attack by a third party. More specifically, to maximize the watermaking effect, we embed the watermark by randomly adding or subtracking a fixed number instead of executing the XORs. Also, to improve the reliability of the watermark detection, we extact the watermark only on some image blocks, which are less sensitive to the compression error. Futhermore, the unrecovered compression errors are further detected by the Hypothesis testing. The illegal signalture attack of a third party is also protected by using some probabilistic decisions of the MSE between the orignal image and the signed image. Experimental results show that the peroposed algorithm is more robust to the quantization errors and illegal signature attack by a third party.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.8
• /
• pp.4285-4299
• /
• 2019

Adversarial attack is a technique that causes a malfunction of classification models by adding noise that cannot be distinguished by humans, which poses a threat to a deep learning model. In this paper, we propose an efficient method to detect adversarial images using Gaussian process regression. Existing deep learning-based adversarial detection methods require numerous adversarial images for their training. The proposed method overcomes this problem by performing classification based on the statistical features of adversarial images and clean images that are extracted by Gaussian process regression with a small number of images. This technique can determine whether the input image is an adversarial image by applying Gaussian process regression based on the intermediate output value of the classification model. Experimental results show that the proposed method achieves higher detection performance than the other deep learning-based adversarial detection methods for powerful attacks. In particular, the Gaussian process regression-based detector shows better detection performance than the baseline models for most attacks in the case with fewer adversarial examples.

• ETRI Journal
• /
• v.31 no.5
• /
• pp.628-630
• /
• 2009

Recently, Ratha and others proposed a cancelable biometrics scheme which transforms an original fingerprint template into a new one using a noninvertible transformation. However, we show that the original template is recovered by a dictionary attack if two transformed templates originating from it are revealed. In our attack, we simulate the transformation and construct a set of possible pre-images for each transformed template. Then, we find the correct pre-image by computing the intersection of these sets. We present an algorithm implementing this idea as well as successful experimental results.

• Journal of Institute of Convergence Technology
• /
• v.4 no.1
• /
• pp.37-41
• /
• 2014

For the analysis of lifting surface at high angle of attack, a Nonlinear Vortex Lattice Method(NLVLM) was used. The NLVLM is intented to compute the interactions between lifting surfaces and separated vertical flow. The lifting surfaces are represented by a lattice of discrete vortex rings. And wakes are represented by families of non-lintersecting, semi-infinite vortex line segments. The image method also used to analyze the ground effect. It is found that vortex lines separated from lifting surfaces represent the separated flows successfully. Although the present method is applied for the rectangular wing and delta wing, extensions can be possible for the arbitrary lifting surfaces. The Present results show good agreement with experimental data.

• Korean journal of communication and information
• /
• v.29
• /
• pp.189-231
• /
• 2005

This study examined image restoration strategies of political parties regarding the Presidential impeachment In Korea. The analytic framework of this study was Benoit's rhetorical theory of image restoration strategies. Results showed that both Hanara Party and Minju Party used attack accuser as a major image restoration strategy. However, Hanara Party changed their strategies from attack accuser to future oriented. Thus, Hanara Party's image restoration strategy successfully applied Coombs theory, which future oriented strategy was good for an organization of the high responsibility in the context of crisis situation.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.57 no.12
• /
• pp.2330-2335
• /
• 2008

In this paper, we proposed a robust watermark scheme for image based on the discrete wavelet transform. In the first, we use the triplet method for embedding and extraction of watermark. Triplet method uses relationship among the discrete wavelet coefficients at the same position in the three detail subband of image. In the second, we use barcode Pattern Image for watermark scheme. By using barcode Pattern Images, a user is able to make many kinds of own watermark. In the third, the proposed method for image watermark provides the invisiblility which the other user can not discriminate watermark from watermarked image. The advantage of the proposed method is visibility when we extracted watermark from the image. The corresponding experimental results show that the proposed method has performance of a robustness from the noise attack.

• Journal of the Korean Society of Visualization
• /
• v.3 no.2
• /
• pp.63-70
• /
• 2005

Birds and insects flap their wings to fly in the air and they can change their wing motions to do steering and maneuvering. Therefore, we created various wing motions with the parameters which affected flapping motion and evaluated the aerodynamic characteristics about those cases in this study. As the wing rotational velocity was fast and the rotational timing was advanced, the measured aerodynamic forces showed drastic increase near the end of stroke. The mean lift coefficient was increased until angle of attack of $50^{\circ}$ and showed the maximum value of 1.0. The maximum mean lift to drag ratio took place at angle of attack of $20^{\circ}$. Flow fields were also visualized around the wing using particle image velocimetry (PIV). From the flow visualization, leading-edge vortex was not shed at mid-stroke until angle of attack of $50^{\circ}$. But it was begun to shed at angle of attack of $60^{\circ}$.

• International Journal of Internet, Broadcasting and Communication
• /
• v.10 no.3
• /
• pp.11-25
• /
• 2018

While being believed that decrypting any RSA ciphertext is as hard as factorizing the RSA modulus, it was also shown that, if additional information is available, breaking the RSA cryptosystem may be much easier than factoring. For example, Coppersmith showed that, given the 1/2 fraction of the least or the most significant bits of one of two RSA primes, one can factorize the RSA modulus very efficiently, using the lattice-based technique. More recently, introducing the so called cold boot attack, Halderman et al. showed that one can recover cryptographic keys from a decayed DRAM image. And, following up this result, Heninger and Shacham presented a polynomial-time attack which, given 0.27-fraction of the RSA private key of the form (p, q, d, $d_p$, $d_q$), can recover the whole key, provided that the given bits are uniformly distributed. And, based on the work of Heninger and Shacham, this paper presents a different approach for recovering RSA private key bits from decayed key information, under the assumption that some random portion of the private key bits is known. More precisely, we present the algorithm of recovering RSA private key bits from erased key material and elaborate the formula of describing the number of partially-recovered RSA private key candidates in terms of the given erasure rate. Then, the result is justified by some extensive experiments.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.12 no.11
• /
• pp.1980-1986
• /
• 2008

A digital watermarking technique used as a protection and certifying mechanism of copyrighted creations including music, still images, and videos in terms of lading any loss in data, reproduction and pursuit. This study suggests using a selected geometric invariant point through the whole processing procedure based on the invariant point so that it will be robust in a geometric transformation attack. The introduced algorithm here is based on a watershed splitting method in order to make medical images strong against RST transformation and other processing. This algorithm also proved that is has robustness against not only RST attack, but also JPEG compression attack and filtering attack.