• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.05a
• /
• pp.219-222
• /
• 2021

인공지능 기술은 모든 분야에서 혁신을 이뤄내고 있다. 이와 동시에 인공지능 모델에 대한 여러 보안적인 문제점이 야기되고 있다. 그 중 대표적인 문제는 많은 인적/물적 자원을 통해 개발한 모델을 악의적인 사용자가 탈취하는 것이다. 모델 탈취가 발생할 경우, 경제적인 문제뿐만 아니라 모델 자체의 취약성을 드러낼 수 있다. 현재 많은 연구가 쿼리를 통해 얻는 모델의 입력과 출력을 분석하여 모델의 의사경계면 또는 모델의 기능성을 탈취하고 있다. 하지만 쿼리 기반의 탈취 공격은 획득할 수 있는 정보가 제한적이기 때문에 완벽한 탈취가 어렵다. 이에 따라 딥러닝 모델 연산 과정에서 데이터 스니핑 또는 캐시 부채널 공격을 통해 추가적인 정보 또는 완전한 모델을 탈취하려는 연구가 진행되고 있다. 본 논문에서는 최근 연구 동향과 쿼리 기반 공격과의 차이점을 분석하고 연구한다.

• International Journal of Computer Science & Network Security
• /
• v.23 no.5
• /
• pp.53-64
• /
• 2023

Early detection of endometrial carcinoma in uterus is essential for effective treatment. Endometrial carcinoma is the worst kind of endometrium cancer among the others since it is considerably more likely to affect the additional parts of the body if not detected and treated early. Non-invasive medical computer vision, also known as medical image processing, is becoming increasingly essential in the clinical diagnosis of various diseases. Such techniques provide a tool for automatic image processing, allowing for an accurate and timely assessment of the lesion. One of the most difficult aspects of developing an effective automatic categorization system is the absence of huge datasets. Using image processing and deep learning, this article presented an artificial endometrium cancer diagnosis system. The processes in this study include gathering a dermoscopy images from the database, preprocessing, segmentation using hybrid Fuzzy C-Means (FCM) and optimizing the weights using the Whale Optimization Algorithm (WOA). The characteristics of the damaged endometrium cells are retrieved using the feature extraction approach after the Magnetic Resonance pictures have been segmented. The collected characteristics are classified using a deep learning-based methodology called Long Short-Term Memory (LSTM) and Bi-directional LSTM classifiers. After using the publicly accessible data set, suggested classifiers obtain an accuracy of 97% and segmentation accuracy of 93%.

• The Bulletin of The Korean Astronomical Society
• /
• v.45 no.1
• /
• pp.60.3-61
• /
• 2020

We perform disk-halo decomposition of the Large Magellanic Cloud (LMC) using a novel HI velocity field extraction method, aimed at better deriving its HI kinematics and thus the dark matter density profile. For this, we use two newly developed galaxy kinematic analysis tools, BAYGAUD and 2DBAT which have been used for the kinematic analysis of resolved galaxies from Australian Square Kilometre Array (ASKAP) observations like WALLABY which is an all-sky HI galaxy survey in southern sky. By applying BAYGAUD to the combined HI data cube of the LMC taken with the Australia Telescope Compact Array (ATCA) and Parkes radio telescopes, we decompose all the line-of-sight velocity profiles into an optimal number of Gaussian components based on Bayesian MCMC techniques. From this, we disentangle turbulent non-circular gas motions from the overall rotation of the galaxy. We then derive the rotation curve of the LMC by applying 2DBAT to the separated circular motions. The rotation curve reflecting the total kinematics of the LMC, dark and baryonic matters is then be combined with the mass models of baryons, mainly stellar and gaseous components in order to examine the dark matter distribution. Here, we present the analysis of the extracted HI gas maps, rotation curve, and J, H and K-band surface photometry of the LMC.

• Communications for Statistical Applications and Methods
• /
• v.30 no.6
• /
• pp.589-603
• /
• 2023

Deep generative models target to infer the underlying true data distribution, and it leads to a huge success in generating fake-but-realistic data. Regarding such a perspective, the data attributes can be a crucial factor in the data generation process since non-existent counterfactual samples can be generated by altering certain factors. For example, we can generate new portrait images by flipping the gender attribute or altering the hair color attributes. This paper proposes counterfactual disentangled variational autoencoder generative adversarial networks (CDVAE-GAN), specialized for data attribute level counterfactual data generation. The structure of the proposed CDVAE-GAN consists of variational autoencoders and generative adversarial networks. Specifically, we adopt a Gaussian variational autoencoder to extract low-dimensional disentangled data features and auxiliary Bernoulli latent variables to model the data attributes separately. Also, we utilize a generative adversarial network to generate data with high fidelity. By enjoying the benefits of the variational autoencoder with the additional Bernoulli latent variables and the generative adversarial network, the proposed CDVAE-GAN can control the data attributes, and it enables producing counterfactual data. Our experimental result on the CelebA dataset qualitatively shows that the generated samples from CDVAE-GAN are realistic. Also, the quantitative results support that the proposed model can produce data that can deceive other machine learning classifiers with the altered data attributes.

• International Journal of Computer Science & Network Security
• /
• v.24 no.2
• /
• pp.89-94
• /
• 2024

Indian subcontinent is a birthplace of multilingual people where documents such as job application form, passport, number plate identification, and so forth is composed of text contents written in different languages/scripts. These scripts may be in the form of different indic numerals in a single document page. Due to this reason, building a generic recognizer that is capable of recognizing handwritten indic digits written by diverse writers is needed. Also, a lot of work has been done for various non-Indic numerals particularly, in case of Roman, but, in case of Indic digits, the research is limited. Moreover, most of the research focuses with only on MNIST datasets or with only single datasets, either because of time restraints or because the model is tailored to a specific task. In this work, a hybrid model is proposed to recognize all available indic handwritten digit images using the existing benchmark datasets. The proposed method bridges the automatically learnt features of Capsule Network with hand crafted Bag of Feature (BoF) extraction method. Along the way, we analyze (1) the successes (2) explore whether this method will perform well on more difficult conditions i.e. noise, color, affine transformations, intra-class variation, natural scenes. Experimental results show that the hybrid method gives better accuracy in comparison with Capsule Network.

• Journal of Chest Surgery
• /
• v.38 no.1 s.246
• /
• pp.38-43
• /
• 2005

Matrix metalloproteinase-2 (MMP-2) is a class of proteolytic enzymes that digest collagen type IV and other components of the basement membrane. It plays a key role in the local invasion and the formation of distant metastases by various malignant tumors. The aim of this study was to evaluate the activity of MMP-2 and its significance as a prognostic marker in resected stage I non-small cell lung cancer (NSCLC). Material and Method: In this study we obtained fresh-frozen samples of tumor and non-tumor tissues from 34 patients with stage I NSCLC who underwent resection without preoperative radiotherapy or chemotherapy. After the extraction of total protein from tissue samples, MMP-2 activities were assessed by gelatin-substrate-zymography. The activities were divided into the higher or lower groups. Result: The MMP-2 activities were higher in tumor tissues than in non-tumor tissues. The MMP-2 activity of non-tumor tissues in recurrent group was higher than in non-recurrent group (p<0.01). Also the patients with higher MMP-2 activity of non-tumor tissues showed poor 5 year survival (p<0.01). Conclusion: This result indicates that the higher level of MMP-2 activity in the non-tumor tissue is associated with the recurrence and survival after the resection of stage I NSCLC. Therefore, MMP-2 activity in the non-tumor tissue could be used as a potential prognostic marker for the resected stage I-NSCLC.

• Journal of IKEEE
• /
• v.26 no.3
• /
• pp.408-415
• /
• 2022

In this paper, we propose a 3D point cloud reconstruction technique from 2D images using efficient feature map extraction network. The originality of the method proposed in this paper is as follows. First, we use a new feature map extraction network that is about 27% efficient than existing techniques in terms of memory. The proposed network does not reduce the size to the middle of the deep learning network, so important information required for 3D point cloud reconstruction is not lost. We solved the memory increase problem caused by the non-reduced image size by reducing the number of channels and by efficiently configuring the deep learning network to be shallow. Second, by preserving the high-resolution features of the 2D image, the accuracy can be further improved than that of the conventional technique. The feature map extracted from the non-reduced image contains more detailed information than the existing method, which can further improve the reconstruction accuracy of the 3D point cloud. Third, we use a divergence loss that does not require shooting information. The fact that not only the 2D image but also the shooting angle is required for learning, the dataset must contain detailed information and it is a disadvantage that makes it difficult to construct the dataset. In this paper, the accuracy of the reconstruction of the 3D point cloud can be increased by increasing the diversity of information through randomness without additional shooting information. In order to objectively evaluate the performance of the proposed method, using the ShapeNet dataset and using the same method as in the comparative papers, the CD value of the method proposed in this paper is 5.87, the EMD value is 5.81, and the FLOPs value is 2.9G. It was calculated. On the other hand, the lower the CD and EMD values, the better the accuracy of the reconstructed 3D point cloud approaches the original. In addition, the lower the number of FLOPs, the less memory is required for the deep learning network. Therefore, the CD, EMD, and FLOPs performance evaluation results of the proposed method showed about 27% improvement in memory and 6.3% in terms of accuracy compared to the methods in other papers, demonstrating objective performance.

• The korean journal of orthodontics
• /
• v.33 no.1 s.96
• /
• pp.1-9
• /
• 2003

The purpose of this study was to find out and evaluate discrepancies between preorthodontic prediction values and actual postorthodontic values and factors contributing to it in 45 patients(17 male, 28 female) who were diagnosed as skeletal Class III ma)occlusion and received presurgical orthodontic treatment and orthognathic surgery at Yonsei university dental hospital. Lateral cephalograms were analysed at pretreatment(T1), orthodontic Prediction(T2), immediately before surgery(T3) and designated the landmark as coordinates or X and Y axes. The samples were divided according to ALD, upper and lower incisor inclination(Ul to SN, IMPA), COS, extraction, the position of extracted teeth and the statistical significance was tested to find out the factors contributing to the prediction. The results were as follows: 1. Differences between preorthodontic prediction values and actual postorthodontic values(T2-T3) were statistically significant(p<0.05) in the x coordinates of U6mbc, L1x and in y coordinates of U1i, U1x, U6me, U6mbc, L6mbc 2. The accuracy of prediction is relatively higher in horizontal changes compared to vortical changes. 3. The statistical significance(p<0.05) between prediction and actual values is observed more in the landmarks of the maxilla than the mandible. 4. Differences between prediction and actual values of incisor and first molar were statistically significant(p<0.05) according to extraction vs non-extraction, extraction type, ALD in the maxilla and according to ALD, IMPA in the mandible. Discrepancies between preorthodontic prediction values and actual postorthodontic values and factors contributing to the prediction must be considered in treatment planning of Cl III surgical patients to increase the accuracy of prediction. Furthermore future investigations are needed on the prediction of vortical changes.

• Journal of Dental Rehabilitation and Applied Science
• /
• v.25 no.2
• /
• pp.139-155
• /
• 2009

Objective : To assess the biological stability of immediate transmucosal placement of tapered implants into tooth extraction sockets. Material and methods : Following tooth extraction, tapered implants were immediately placed into the sockets. Teeth with evidence of acute periapical pathology were excluded. After implant placement, sutured allowing a non-submerged, transmucosal healing. Standardized radiographs were obtained every visiting from baseline to 32 weeks after implant placment. Changes in depth of the distance from the implant shoulder (IS) and from the alveolar crest (AC) to the bottom of the defect (BD) were assessed. Results : Thirteen patients (10 males and 3 females) were enrolled and followed. They contributed with 15 tapered implants. extraction iste displayed sufficient residual bone volume to allow primary stability of all implants. The mean surgery time was $41{\pm}10.0$ mins. All implants healed uneventfully yielding a survival rate of 100%. Mean ISQ values were relatively stable. Interproximal crestal bone decreased $1.69{\pm}1.2mm$ (mesial), $1.65{\pm}1.2mm$ (distal) from baseline to 32-week follow-up. No statistically significant changes with respect to FMPS, FMBS, PPD and width of KG were observed. Conclusions: Immediate transmucosal implant placement represented a predictable treatment option for the replacement of teeth lost due to reasons including fractures, endodontic failures and caries.

• Archives of Pharmacal Research
• /
• v.26 no.11
• /
• pp.967-973
• /
• 2003

Cytochrome P4502C9(CYP2C9) is largely responsible for terminating anticoagulant effect by hydroxylation of S-warfarin to inactive metabolites. Mutations in the CYP2C9 gene result in the expression of allelic variants, CYP2C9*2 and CYP2C9*3 with reduced enzyme activity compared to wild type CYP2C9 *1. The aim of this study was to assess relationship between requirement of warfarin dose and polymorphism in CYP2C9 in Korean population. Patients on warfarin therapy for longer than 1 year were included from July 1999 to December 2000 and categorized as one of four groups; regular dose non-bleeding, regular dose bleeding, low dose non-bleeding and low dose bleeding. Low dose was defined as less than 10 mg/week for 3 consecutive monthly follow-ups. Bleeding complications included minor and major bleedings. Blood samples were processed for DNA extraction, genotyping and sequencing to detect polymorphism in CYP2C9. Demographic data, warfarin dose per week, prothrombin time (INR), indications and co-morbid diseases were assessed for each group. Total 90 patients on warfarin were evaluated; The low dose group has taken warfarin 7.6$\pm$1.7 mg/week, which was significantly lower than 31.4$\pm$0.9 mg/week in the regular dose group (p<0.0001). The measured INR in the low dose group was similar to that of the regular dose group (2.3$\pm$0.7 vs. 2.3$\pm$0.6, p=0.9). Even though there was a higher possibility of CYP2C9 variation in the low dose group, no polymorphism in CYP2C9 was detected. All patients were homozygous C416 in exon 3 for CYP2C9*2 and A1061 in exon 7 for CYP2C9*3. The DNA sequencing data confirmed the homozygous C416 and A 1061 alleles. In conclusion, polymorphism in CYP2C9 is not a critical factor for assessing warfarin dose requirement and risk of bleeding complications in a Korean population.