• 인터넷정보학회논문지
• /
• 제19권3호
• /
• pp.15-23
• /
• 2018

코드 재사용 공격은 프로그램 메모리상에 존재하는 실행 가능한 코드 조각을 조합하고, 이를 연속적으로 실행함으로써 스택에 직접 코드를 주입하지 않고도 임의의 코드를 실행시킬 수 있는 공격 기법이다. 코드 재사용 공격의 대표적인 종류로는 ROP(Return-Oriented Programming) 공격이 있으며, ROP 공격에 대응하기 위한 여러 방어기법들이 제시되어왔다. 그러나 기존의 방법들은 특정 규칙을 기반으로 공격을 탐지하는 Rule-base 방식을 사용하기 때문에 사전에 정의한 규칙에 해당되지 않는 ROP 공격은 탐지할 수 없다는 한계점이 존재한다. 본 논문에서는 RNN(Recurrent Neural Network)을 사용하여 ROP 공격 코드에 사용되는 명령어 패턴을 학습하고, 이를 통해 ROP 공격을 탐지하는 방법을 소개한다. 또한 정상 코드와 ROP 공격 코드 판별에 대한 False Positive Ratio, False Negative Ratio, Accuracy를 측정함으로써 제안한 방법이 효과적으로 ROP 공격을 탐지함을 보인다.

• 대한의용생체공학회:의공학회지
• /
• 제18권4호
• /
• pp.485-491
• /
• 1997

본 논문은 각성상태에 다른 생리신호와 행위신호 및 주관적 평가의 상관관계에 대하여 나타내었다. Nz와 반응시간은 mKSS level 의 변화와 동일한 경향을 나타내는데 반하여 1분당 눈 깜박임 수는 앞의 두 가지 변수와 다른 경향을 나타내었다. 1분당 눈깜박임 수는 mKSS level 1에서 5까지는 낮은 변화율 갖고 mKSS level 7에서는 높은 변화율을 갖는 반면에 mKSS level 9에서는 이와 반대로 변화율이 급격히 감소한다. 피검자들은 서로다른 1분당 눈깜박임 수(EBR)를 가지나 EBR의 변화율은 비슷하였다. 그러므로 EBR의 변화율을 각성판정지표로 사용할 수 있음을 알 수 있었다. 반응시간 실험 결과로부터mKSS level 5이상부터 작업수행능력이 낮아짐을 알 수 있었고 false positive 와 false negative 가 mKSS level3부터 관찰되었으므로 효과적으로 각성제어를 위하여 mKSS level 3과 5사이에 각성상태를 향상시키기 위한 소리나 향기 등의 자극을 주어야 함을 알 수 있었다.

• Imaging Science in Dentistry
• /
• 제48권2호
• /
• pp.121-125
• /
• 2018

Purpose: Detecting laryngeal cartilages (triticeous and thyroid cartilages) on panoramic radiographs is important because they may be confused with carotid artery calcifications in the bifurcation region, which are a risk factor for stroke. This study assessed the efficiency of panoramic radiography in the diagnosis of calcified laryngeal cartilages using cone-beam computed tomography (CBCT) as the reference standard. Materials and Methods: A total of 312 regions(142 bilateral, 10 left, 18 right) in 170 patients(140 males, 30 females) were examined. Panoramic radiographs were examined by an oral and maxillofacial radiologist with 11 years of experience. CBCT scans were reviewed by 2 other oral and maxillofacial radiologists. The kappa coefficient(${\kappa}$) was calculated to determine the level of intra-observer agreement and to determine the level of agreement between the 2 methods. Diagnostic indicators(sensitivity, specificity, accuracy, and false positive and false negative rates) were also calculated. P values <.05 were considered to indicate statistical significance. Results: Eighty-two images were re-examined to determine the intra-observer agreement level, and the kappa coefficient was calculated as 0.709 (P<.05). Statistically significant and acceptable agreement was found between the panoramic and CBCT images (${\kappa}=0.684$ and P<.05). The sensitivity, specificity, diagnostic accuracy rate, the false positive rate, and the false negative rate of the panoramic radiographs were 85.4%, 83.5%, 84.6%, 16.5%, and 14.6%, respectively. Conclusion: In most cases, calcified laryngeal cartilages could be diagnosed on panoramic radiographs. However, due to variation in the calcifications, diagnosis may be difficult.

• 정보처리학회논문지C
• /
• 제13C권1호
• /
• pp.45-54
• /
• 2006

현재까지 존재하는 무수한 악성 행위에 대응하기 위해서 다양한 기법들이 제안되었다 그러나 현존하는 악성행위 탐지 기법들은 기존의 행위에 대한 변종들과 새로운 형태의 악성행위에 대해서 적시 적절하게 대응하지 못하였고 긍정 오류(false positive)와 틀린 부정(negative false) 등을 해결하지 못한 한계점을 가지고 있다. 위와 같은 문제점을 개선하고자 한다. 여기서는 소스코드의 기본 단위(token)들을 개념화하여 악성행위 탐지에 응용하고자 한다. 악성 코드를 개념 그래프로 정의할 수 있고, 정의된 그래프를 통하여 정규화 표현으로 바꿔서 코드 내 악성행위 유사관계를 비교할 수 있다. 따라서 본 논문에서는, 소스코드를 개념 그래프화하는 방법을 제시하며, 정확한 악성행위 판별을 위한 유사도 측정방안을 제시한다. 실험결과, 향상된 악성 코드 탐지율을 얻었다.

• 디지털콘텐츠학회 논문지
• /
• 제18권2호
• /
• pp.373-382
• /
• 2017

과학기술콘텐츠가 증가함에 따라 과학기술콘텐츠의 효율적인 검색을 지원하는 서비스가 요구되고 있다. 저자의 소속기관명을 키워드로 사용할 경우 한 기관에서 생산된 콘텐츠를 확인할 수 있을 뿐만 아니라 저자, 용어를 키워드로 사용한 검색 결과의 식별율을 향상 시킬 수 있다. 검색 키워드로 사용되는 데이터들의 중의성과 모호성으로 인해 검색 결과에 false negative, false positive가 포함될 수 있으므로 데이터의 식별을 통한 통제는 중요하다. 저자의 소속기관명의 식별을 통한 통제 역시 기관의 이명, 약어 검색을 지원가능하게 하므로 매우 중요하지만 기존의 데이터 식별을 통한 통제에 대한 연구는 저자, 용어에 대한 연구가 주를 이루었다. 본 연구에서는 기관명 식별 알고리즘을 제안하고, 한국과학기술정보연구원에서 보유하고 있는 국내 과학기술콘텐츠들에 대한 데이터를 이용한 실험 결과를 보인다.

• 한국농공학회논문집
• /
• 제61권3호
• /
• pp.55-65
• /
• 2019

Although crack on concrete exists from its early formation, crack requires attention as it affects stiffness of structure and can lead demolition of structure as it grows. Detecting cracks on concrete is needed to take action prior to performance degradation of structure, and deep learning can be utilized for it. In this study, transfer learning, one of the deep learning techniques, was used to detect the crack, as the amount of crack's image data was limited. Pre-trained Inception-v3 was applied as a base model for the transfer learning. Web scrapping was utilized to fetch images of concrete wall with or without crack from web. In the recognition of crack, image post-process including changing size or removing color were applied. In the visualization of crack, source images divided into 30px, 50px or 100px size were used as input data, and different numbers of input data per category were applied for each case. With the results of visualized crack image, false positive and false negative errors were examined. Highest accuracy for the recognizing crack was achieved when the source images were adjusted into 224px size under gray-scale. In visualization, the result using 50 data per category under 100px interval size showed the smallest error. With regard to the false positive error, the best result was obtained using 400 data per category, and regarding to the false negative error, the case using 50 data per category showed the best result.

• Korean Journal of Radiology
• /
• 제21권7호
• /
• pp.891-899
• /
• 2020

Objective: To assess the diagnostic performance of a deep learning-based algorithm for automated detection of acute and chronic rib fractures on whole-body trauma CT. Materials and Methods: We retrospectively identified all whole-body trauma CT scans referred from the emergency department of our hospital from January to December 2018 (n = 511). Scans were categorized as positive (n = 159) or negative (n = 352) for rib fractures according to the clinically approved written CT reports, which served as the index test. The bone kernel series (1.5-mm slice thickness) served as an input for a detection prototype algorithm trained to detect both acute and chronic rib fractures based on a deep convolutional neural network. It had previously been trained on an independent sample from eight other institutions (n = 11455). Results: All CTs except one were successfully processed (510/511). The algorithm achieved a sensitivity of 87.4% and specificity of 91.5% on a per-examination level [per CT scan: rib fracture(s): yes/no]. There were 0.16 false-positives per examination (= 81/510). On a per-finding level, there were 587 true-positive findings (sensitivity: 65.7%) and 307 false-negatives. Furthermore, 97 true rib fractures were detected that were not mentioned in the written CT reports. A major factor associated with correct detection was displacement. Conclusion: We found good performance of a deep learning-based prototype algorithm detecting rib fractures on trauma CT on a per-examination level at a low rate of false-positives per case. A potential area for clinical application is its use as a screening tool to avoid false-negative radiology reports.

• Journal of the Optical Society of Korea
• /
• 제14권4호
• /
• pp.451-457
• /
• 2010

Increasing demands on the safety of public train services have led to the development of various types of security monitoring systems. Most of the surveillance systems are focused on the estimation of crowd level in the platform, thereby yielding too many false alarms. In this paper, we present a novel security monitoring system to detect critically dangerous situations such as when a passenger falls from the station platform, or when a passenger walks on the rail tracks. The method is composed of two stages of detecting dangerous situations. Objects falling over to the dangerous zone are detected by motion tracking. 3D depth information retrieved by the stereo vision is used to confirm fallen events. Experimental results show that virtually no error of either false positive or false negative is found while providing highly reliable detection performance. Since stereo matching is performed on a local image only when potentially dangerous situations are found; real-time operation is feasible without using dedicated hardware.

• 한국정보시스템학회지:정보시스템연구
• /
• 제15권4호
• /
• pp.211-224
• /
• 2006

This study investigates the application of data mining techniques such as artificial neural networks, rough sets, and induction teaming to the intrusion detection systems. To maximize the effectiveness of data mining for intrusion detection systems, we introduced the asymmetric costs with false positive errors and false negative errors. And we present a method for intrusion detection systems to utilize the asymmetric costs of errors in data mining. The results of our empirical experiment show our intrusion detection model provides high accuracy in intrusion detection. In addition the approach using the asymmetric costs of errors in rough sets and neural networks is effective according to the change of threshold value. We found the threshold has most important role of intrusion detection model for decreasing the costs, which result from false negative errors.

• Journal of Chest Surgery
• /
• 제1권1호
• /
• pp.75-80
• /
• 1968

This is a case report of traumatic arterial injuries with false aneurysm & arterio-venous fistula treated surgically at National Medical Center. 3 cases were A-V fistula and 2 cases only false aneurysm. Physiological disturbance were produced by only arteriovenous fistula; In one case ulceration of mid. 1/3 tibia due to diminished arterial flow and in 2 cases left ventricular hypertrophy, in which cases Bramhan`s sign were positive. Removing out the fistulous lesions and aneurysm, all of the arterial continuities has been reconstructed by means of end to end anastomosis, Dacron graft and vein graft, veins were managed by ligations of both ends in two cases and end to end anostomosis in one case. Immediate post operative results were good, and two cases were followed for 10 months.