• Journal of IKEEE
• /
• v.28 no.1
• /
• pp.6-13
• /
• 2024

This paper proposes an approach to detect abnormal data in automotive controller area network (CAN) using an unsupervised learning model, i.e. autoencoder and Gaussian kernel density estimation function. The proposed autoencoder model is trained with only message ID of CAN data frames. Afterwards, by employing the Gaussian kernel density estimation function, it effectively detects abnormal data based on the trained model characterized by the optimally determined number of frames and a loss threshold. It was verified and evaluated using four types of attack data, i.e. DoS attacks, gear spoofing attacks, RPM spoofing attacks, and fuzzy attacks. Compared with conventional unsupervised learning-based models, it has achieved over 99% detection performance across all evaluation metrics.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.8
• /
• pp.85-92
• /
• 2024

In this paper presents a comparative analysis of the performance between the latest object detection model, YOLOv10, and its previous versions. YOLOv10 introduces NMS-Free training, an enhanced model architecture, and an efficiency-centric design, resulting in outstanding performance. Experimental results using the COCO dataset demonstrate that YOLOv10-N maintains high accuracy of 39.5% and low latency of 1.84ms, despite having only 2.3M parameters and 6.7G floating-point operations (FLOPs). The key performance metrics used include the number of model parameters, FLOPs, average precision (AP), and latency. The analysis confirms the effectiveness of YOLOv10 as a real-time object detection model across various applications. Future research directions include testing on diverse datasets, further model optimization, and expanding application scenarios. These efforts aim to further enhance YOLOv10's versatility and efficiency.

• Science of Emotion and Sensibility
• /
• v.21 no.1
• /
• pp.177-186
• /
• 2018

The mechanism of emotion is complex and influenced by a variety of factors, so that it is crucial to analyze emotion in broad and diversified perspectives. In this study, we classified neutral and negative emotions(sadness, fear, surprise) using arousal evaluation, which is one of the psychological evaluation scales, as well as physiological signals. We have not only revealed the difference between physiological signals coupled to the emotions, but also assessed how accurate these emotions can be classified by our emotional recognizer based on neural network algorithm. A total of 146 participants(mean age $20.1{\pm}4.0$, male 41%) were emotionally stimulated while their physiological signals of the electrocardiogram, blood flow, and dermal activity were recorded. In addition, the participants evaluated their psychological states on the emotional rating scale in response to the emotional stimuli. Heart rate(HR), standard deviation(SDNN), blood flow(BVP), pulse wave transmission time(PTT), skin conduction level(SCL) and skin conduction response(SCR) were calculated before and after the emotional stimulation. As a result, the difference between physiological responses was verified corresponding to the emotions, and the highest emotion classification performance of 86.9% was obtained using the combined analysis of arousal and physiological features. This study suggests that negative emotion can be categorized by psychological and physiological evaluation along with the application of machine learning algorithm, which can contribute to the science and technology of detecting human emotion.

• Economic and Environmental Geology
• /
• v.36 no.2
• /
• pp.123-131
• /
• 2003

Geophysical surveys(electrical resistivity, self-potential, seismic refraction, GPR) were conducted to investigate the physical properties of the subsurface, and to delineate the flow channel of leachate from a AMD(acid mine drainage), buried rock wastes and tailings, and drainage pipes at an abandoned mine(Kwangyang mine). Especially in rainy season the sites appear to be abundant in AMD leachate, characterized by electrical conductivities of 0.98-1.10 ms/S. Electrical resistivity sections indicate that the leachate flows running in two directions at southern part rise up through the narrow fracture zones at the central part and contaminates the surrounding soil and stream. Such schematic features at the anomalous zone are well correlated with negative peaks in self-potential data, the limited penetration depth in GPR data and low velocity zone in seismic refraction data. Shallow high-resistivity zone is associated with the buried rock wastes which cause the diffractions in GPR image. In addition, the events at depth of approximately 1-1.25 m in GPR sections must be the metal pipes through which AMD is drained off to the inner bay.

• Korean Journal of Remote Sensing
• /
• v.21 no.4
• /
• pp.273-285
• /
• 2005

This study has investigated the suitable wavelength for detecting biomass burning aerosols. We have performed the analysis of the wavelength at 380nm in near-UV, 400nm, 412nm, 460nm, and 490nm in visible, and 2100nm in shortwave infrared regions from the Global Imager measurements. It is well known that the UV bands have the advantage of the aerosols retrieval due to the low surface reflectance and a weak effect of Bidirectional Reflectivity Distribution Function. However, the pure surface reflectances of shortwave visible bands, except 412nm, are as low as that of 380nm in near-UV over northeast Asia. In order to detect the aerosol signal, we have retrieved the aerosol reflectance as a function of wavelength based on the surface reflectivity contrast method for the period of May 2003. It is interesting that the retrieved aerosol reflectance with 460nm is slightly more sensitive than that with 380nm. Additionally, we have applied the TOMS aerosol index method to determine the best pair for biomass burning aerosols and found that the pair of 380 and 460nm results in the best signal for retrieving aerosols.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.18 no.5
• /
• pp.538-547
• /
• 2015

The goal of this study is to develop an algorithm to propose optimal deployment of detection sensor nodes in the target area, based on a performance surface, which represents detection performance of active and passive acoustic sonar systems. The performance surface of the active detection system is calculated from the azimuthal average of maximum detection ranges, which is estimated with a transmission loss and a reverberation level predicted using ray-based theories. The performance surface of the passive system is calculated using the transmission loss model based on a parabolic equation. The optimization of deployment configurations is then performed by a hybrid method of a virtual force algorithm and a particle swarm optimization. Finally, the effectiveness of deployment configurations is analyzed and discussed with the simulation results obtained using the algorithm proposed in this paper.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2012.05a
• /
• pp.256-256
• /
• 2012

2011년 태국 차오프라야 강 유역($160,813km^2$)에서 발생한 홍수에 의해 많은 피해가 발생했다. 태국 홍수는 2011년 7월 말부터 3개월간 내린 집중호우로 중부지방에 50년 만에 최악의 자연재해를 맞이하였다. 태국 북쪽 지역에서 난 강과 핑 강의 범람을 시작으로 태국 중앙 지역을 흐르는 차오프라야 강의 수위는 상류의 홍수가 하류로 내려옴에 따라 범람하여 수도 방콕까지 침수되었다. 본 연구에서는 홍수범람시 시공간적 침수상황이 파악 가능한 Terra MODIS (Moderate Resolution Imaging Spectroradiometer) 영상을 이용하여 태국 차오프라야 강 유역의 홍수에 의한 침수지역을 추정하고자 하였다. 2011년 7월 29일에서 2012년 1월 9일까지의 500 m 해상도인 MODIS product MOD09 (Surface Reflectance) 8일 합성 영상을 수집하고 식생지수 (EVI; Enhanced Vegetation Index), 지표수분지수 (LSWI; Land Surface Water Index))와 DVEL지수 (Difference Value between EVI and LSWI)를 이용하여 홍수범람 지역과 수역관련지역을 정보화 기법을 제시하였다. 본 연구의 결과는 홍수 범람지역의 자료를 정보화하고 그 결과를 정량적으로 제시하는 방법으로 활용될 수 있으며, MODIS 자료의 이용은 시공간적 하천 홍수범람지역 탐지의 가능성을 알 수 있었다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2015.10a
• /
• pp.91-94
• /
• 2015

최근 들어 공용 클라우드 서비스의 이용이 크게 증가하였다. 클라우스 서비스는 관리비용이 저렴하고 다양한 디바이스로 접근이 가능한 장점이 있는 반면, 데이터 접근을 통제하기 힘들어 내부정보가 유출되는 등의 보안 위협이 공존한다. 따라서 본 논문은 클라우드 스토리지의 기밀문서 유출 방지를 위해 사후 감시방법을 연구한다. 문서 내 포함된 보안 키워드 개수를 통해 보안지표를 계산하고 이 지표로 기밀문서 여부를 판단한다. 또한, 관리자용 대시보드를 통해 기밀문서가 탐지 되었을 때 백업/삭제 등의 사후처리 기능을 제공한다. 본 감시방법을 통해 클라우드 스토리지 내 기밀문서가 업로드 되는 것을 감시함으로써 기밀문서 유출 위협을 효과적으로 줄일 수 있다.

• Journal of the Earthquake Engineering Society of Korea
• /
• v.7 no.4
• /
• pp.89-95
• /
• 2003

In this paper, an algorithm to locate and size damage in a complex truss structure using the time domain response is presented. Sampled response data for specific time interval is spatially expanded over the structure to obtain the mean train energy for each element of the structure. The mean strain energy for each element is, in turn, used to build a damage index that represents the ratio of the stiffness parameter of the pre-damaged to the post-damaged structure. The validity of the methodology is demonstrated using data from a numerical example of a space truss structure with simulated damage. Also in the example, the effects of noisy data on the proposed algorithm are examined by adding random noised to the response data.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.21 no.12
• /
• pp.320-330
• /
• 2020

Fire is difficult to achieve good performance in image detection using deep learning because of its high irregularity. In particular, there is little data on fire detection in underground utility facilities, which have poor light conditions and many objects similar to fire. These make fire detection challenging and cause low performance of deep learning models. Therefore, this study proposed a fire detection model using deep learning and estimated the performance of the model. The proposed model was designed using a combination of a basic convolutional neural network, Inception block of GoogleNet, and Skip connection of ResNet to optimize the deep learning model for fire detection under underground utility facilities. In addition, a training technique for the model was proposed. To examine the effectiveness of the method, the trained model was applied to fire images, which included fire and non-fire (which can be misunderstood as a fire) objects under the underground facilities or similar conditions, and results were analyzed. Metrics, such as precision and recall from deep learning models of other studies, were compared with those of the proposed model to estimate the model performance qualitatively. The results showed that the proposed model has high precision and recall for fire detection under low light intensity and both low erroneous and missing detection capabilities for things similar to fire.