• Journal of Internet Computing and Services
• /
• v.22 no.4
• /
• pp.1-11
• /
• 2021

As the Internet developed, various and complex cyber attacks began to emerge. Various detection systems were used outside the network to defend against attacks, but systems and studies to detect attackers inside were remarkably rare, causing great problems because they could not detect attackers inside. To solve this problem, studies on the lateral movement detection system that tracks and detects the attacker's movements have begun to emerge. Especially, the method of using the Remote Desktop Protocol (RDP) is simple but shows very good results. Nevertheless, previous studies did not consider the effects and relationships of each logon host itself, and the features presented also provided very low results in some models. There was also a problem that the model could not explain why it predicts that way, which resulted in reliability and robustness problems of the model. To address this problem, this study proposes an interpretable RDP-based lateral movement detection system using page rank algorithm and SHAP(Shapley Additive Explanations). Using page rank algorithms and various statistical techniques, we create features that can be used in various models and we provide explanations for model prediction using SHAP. In this study, we generated features that show higher performance in most models than previous studies and explained them using SHAP.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.48 no.2
• /
• pp.64-70
• /
• 2011

Portable artificial electronic nose (E-nose) system suffers from noisy fluctuation in surroundings such as temperature, vapor concentration, and gas flow, because its measuring condition is not controled precisely as in the laboratory. It is important to develop a simple and robust vapor recognition technique applicable to this uncontrolled measurement, especially for the portable measuring and diagnostic system which are expanding its area with the improvements in micro bio sensor technology. This study used a PDA-based portable E-nose to collect the uncontrolled vapor measurement signals, and applied the image matching algorithm developed in the previous study on the measured signal to verify its robustness and improved accuracy in portable vapor recognition. The results showed not only its consistent performance under noisy fluctuation in the portable measurement signal, but also an advanced recognition accuracy for 2 similar vapor species which have been hard to discriminate with the conventional maximum sensitivity feature extraction method. The proposed method can be easily applied to the data processing of the ubiquitous sensor network (USN) which are usually exposed to various operating conditions. Furthermore, it will greatly help to realize portable medical diagnostic and environment monitoring system with its robust performance and high accuracy.

• The KIPS Transactions:PartB
• /
• v.12B no.1 s.97
• /
• pp.9-16
• /
• 2005

This paper focuses on the study of the face authentication system and the robustness of fact authentication methods under illumination changes. Four different face authentication methods are tried. These methods are as fellows; PCA(Principal Component Analysis), GMM(Gaussian Mixture Modeis), 1D HMM(1 Dimensional Hidden Markov Models), Pseudo 2D HMM(Pseudo 2 Dimensional Hidden Markov Models). Experiment results involving an artificial illumination change to fate images are compared with each other. Face feature vector extraction based on the 2D DCT(2 Dimensional Discrete Cosine Transform) if used. Experiments to evaluate the above four different fate authentication methods are carried out on the ORL(Olivetti Research Laboratory) face database. Experiment results show the EER(Equal Error Rate) performance degrade in ail occasions for the varying ${\delta}$. For the non illumination changes, Pseudo 2D HMM is $2.54{\%}$,1D HMM is $3.18{\%}$, PCA is $11.7{\%}$, GMM is $13.38{\%}$. The 1D HMM have the bettor performance than PCA where there is no illumination changes. But the 1D HMM have worse performance than PCA where there is large illumination changes(${\delta}{\geq}40$). For the Pseudo 2D HMM, The best EER performance is observed regardless of the illumination changes.

• The KIPS Transactions:PartB
• /
• v.18B no.3
• /
• pp.147-156
• /
• 2011

This paper proposes a 3-stage (preprocessing, feature extraction, and classification) fault detection and classification algorithm for induction motors. In the first stage, a low-pass filter is used to remove noise components in the fault signal. In the second stage, a discrete cosine transform (DCT) and a statistical method are used to extract features of the fault signal. Finally, a back propagation neural network (BPNN) method is applied to classify the fault signal. To evaluate the performance of the proposed algorithm, we used one second long normal/abnormal vibration signals of an induction motor sampled at 8kHz. Experimental results showed that the proposed algorithm achieves about 100% accuracy in fault classification, and it provides 50% improved accuracy when compared to the existing fault detection algorithm using a cross-covariance method. In a real-world data acquisition environment, unnecessary noise components are usually included to the real signal. Thus, we conducted an additional simulation to evaluate how well the proposed algorithm classifies the fault signals in a circumstance where a white Gaussian noise is inserted into the fault signals. The simulation results showed that the proposed algorithm achieves over 98% accuracy in fault classification. Moreover, we developed a testbed system including a TI's DSP (digital signal processor) to implement and verify the functionality of the proposed algorithm.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2008.05a
• /
• pp.1077-1081
• /
• 2008

Soil and Water Assessment Tool(SWAT) 모형은 DEM(Digital Elevation Model)을 사용하여 지형인자를 추출하고 이를 바탕으로 수문 및 수질 모의가 이루어진다. 지형인자의 추출시 DEM 격자크기에 따라 상이한 결과를 초래할 수 있다. 그리하여 정확한 수문 및 수질 모델링에 있어 가능한 고해상도의 DEM을 사용하도록 권장하고 있다. 그러나 넓은 유역에서의 적용시 고해상도 DEM 사용에 따른 컴퓨터 처리 용량과 프로그램 실행 시 소요되는 시간상의 문제는 그 효율성에 있어서 문제시될 수 있다. 그리하여 본 연구에서는 소양강댐, 임하댐 유역을 대상으로 SWAT 모형에서 저해상도 DEM 사용으로 고해상도 DEM의 지형인자를 추출하여 자동 입력될 수 있는 모듈을 개발 적용하였다. 본 연구의 결과 소양강댐 유역을 대상으로 격자크기 20m DEM과 100m DEM을 사용하였을 때 연평균 유사량이 83.8%의 큰 차이가 발생한 반면 격자크기의 20m DEM과 본 모듈을 적용하여 20m DEM의 지형인자로 자동 보정된 100m DEM을 사용하였을 때의 연평균 유사량이 4.4%로 차이가 상당히 줄어든 것을 볼 수 있었다. 임하댐 유역의 경우는 격자크기 10m DEM과 100m DEM을 사용하였을 때 연평균 유사량이 43.4% 큰 차이가 발생하였다. 반면 격자크기 10m DEM과 본 모듈을 적용하여 10m DEM의 지형인자로 자동 보정된 100m DEM을 사용하였을 때의 연평균 유사량이 0.3%로 차이가 크게 줄어든 것을 확인 할 수 있었다. 또한 본 모듈의 검정을 위해 소양강댐 유역의 지형 자료와 유사한 충주댐 유역을 대상으로 본 모듈을 적용하여 검정을 실시하였다. 그 결과 연간 평균 유사량이 격자크기 20m와 100m의 DEM을 이용하였을 때 98.7%의 큰 차이가 발생한 반면 격자크기 20m와 본 모듈을 적용하여 보정된 경사도 값의 100m DEM을 사용하였을 때 20.7%로 차이가 크게 줄어든 것을 볼 수 있었다. 그리하여 본 연구의 결과를 통해 SWAT 모형에서의 개선된 지형인자 추출 방식을 사용하여 저해상도의 DEM 사용으로 고해상도 DEM 사용의 효과를 볼 수 있을 것이고 이로 인해 넓은 유역에서 저해상도 DEM 사용으로 컴퓨터 사용용량과 프로그램 지연 시간을 줄일 수 있을 것으로 판단된다. 향후 여러 유역을 대상으로 보정, 검정하여 보다 정확하고 통합적으로 적용될 수 있는 모듈의 개선이 필요할 것으로 사료된다.

• Economic and Environmental Geology
• /
• v.38 no.2 s.171
• /
• pp.101-111
• /
• 2005

The objectives of this study we to assess the extent and degree of environmental contamination and to draw general conclusions on the fate of toxic elements derived from mining activities in Korea. 인t abandoned mines with four base-metal mines and four Au-Ag mines were selected and the results of environmental surveys in those areas were discussed. In the base-metal mining areas, the Sambo Pb-Zn-barite, the Shinyemi Pb-Zn-Fe, the Geodo Cu-Fe and the Shiheung Cu-Pb-Zn mine, significant levels of Cd, Cu, Pb and Zn were found in mine dump soils developed over mine waste materials, tailings and slag. Furthermore, agricultural soils, stream sediments and stream water near the mines were severely contaminated by the metals mainly due to the continuing dispersion downstream and downslope from the sites, which was controlled by the feature of geography, prevailing wind directions and the distance from the mine. In e Au-Ag mining areas, the Kubong, the Samkwang, the Keumwang and the Kilkok mines, elevated levels of As, Cd, Cu, Pb and Zn were found in tailings and mine dump soils. These levels may have caused increased concentrations of those elements in stream sediments and waters due to direct dis-charge downstream from tailings and mine dumps. In the Au-Ag mines, As would be the most characteristic contaminant in the nearby environment. Arsenic and heavy metals were found to be mainly associated with sulfide gangue minerals, and mobility of these metals would be enhanced by the effect of oxidation. According to sequential extraction of metals in soils, most heavy metals were identified as non-residual chemical forms, and those are very susceptible to the change of ambient conditions of a nearby environment. As application of pollution index (PI), giving data on multi-element contamination in soils, over 1.0 value of the PI was found in soils sampled at and around the mining areas.

• Journal of Intelligence and Information Systems
• /
• v.26 no.2
• /
• pp.1-25
• /
• 2020

In this paper, we suggest an application system architecture which provides accurate, fast and efficient automatic gasometer reading function. The system captures gasometer image using mobile device camera, transmits the image to a cloud server on top of private LTE network, and analyzes the image to extract character information of device ID and gas usage amount by selective optical character recognition based on deep learning technology. In general, there are many types of character in an image and optical character recognition technology extracts all character information in an image. But some applications need to ignore non-of-interest types of character and only have to focus on some specific types of characters. For an example of the application, automatic gasometer reading system only need to extract device ID and gas usage amount character information from gasometer images to send bill to users. Non-of-interest character strings, such as device type, manufacturer, manufacturing date, specification and etc., are not valuable information to the application. Thus, the application have to analyze point of interest region and specific types of characters to extract valuable information only. We adopted CNN (Convolutional Neural Network) based object detection and CRNN (Convolutional Recurrent Neural Network) technology for selective optical character recognition which only analyze point of interest region for selective character information extraction. We build up 3 neural networks for the application system. The first is a convolutional neural network which detects point of interest region of gas usage amount and device ID information character strings, the second is another convolutional neural network which transforms spatial information of point of interest region to spatial sequential feature vectors, and the third is bi-directional long short term memory network which converts spatial sequential information to character strings using time-series analysis mapping from feature vectors to character strings. In this research, point of interest character strings are device ID and gas usage amount. Device ID consists of 12 arabic character strings and gas usage amount consists of 4 ~ 5 arabic character strings. All system components are implemented in Amazon Web Service Cloud with Intel Zeon E5-2686 v4 CPU and NVidia TESLA V100 GPU. The system architecture adopts master-lave processing structure for efficient and fast parallel processing coping with about 700,000 requests per day. Mobile device captures gasometer image and transmits to master process in AWS cloud. Master process runs on Intel Zeon CPU and pushes reading request from mobile device to an input queue with FIFO (First In First Out) structure. Slave process consists of 3 types of deep neural networks which conduct character recognition process and runs on NVidia GPU module. Slave process is always polling the input queue to get recognition request. If there are some requests from master process in the input queue, slave process converts the image in the input queue to device ID character string, gas usage amount character string and position information of the strings, returns the information to output queue, and switch to idle mode to poll the input queue. Master process gets final information form the output queue and delivers the information to the mobile device. We used total 27,120 gasometer images for training, validation and testing of 3 types of deep neural network. 22,985 images were used for training and validation, 4,135 images were used for testing. We randomly splitted 22,985 images with 8:2 ratio for training and validation respectively for each training epoch. 4,135 test image were categorized into 5 types (Normal, noise, reflex, scale and slant). Normal data is clean image data, noise means image with noise signal, relfex means image with light reflection in gasometer region, scale means images with small object size due to long-distance capturing and slant means images which is not horizontally flat. Final character string recognition accuracies for device ID and gas usage amount of normal data are 0.960 and 0.864 respectively.

• The Journal of the Petrological Society of Korea
• /
• v.17 no.1
• /
• pp.16-35
• /
• 2008

We study metamorphism of metasedimetary rocks and origin and evolution of leucogranite form Samcheok area, northeastern Yeongnam massif, South Korea. Metamorphic rocks in this area are composed of metasedimentary migmatite, biotite granitic gneiss and leucogranite. Metasedimentary rocks, which refer to major element feature of siliclastic sediment, are divided into two metamorphic zones based on mineral assemblages, garnet and sillimanite zones. According to petrogenetic grid of mineral assemblages, metamorhpic P-T conditions are $740{\sim}800^{\circ}C$ at $4.8{\sim}5.8\;kbar$ in the garnet zone and $640-760^{\circ}C$ at 2.5-4.5kbar in sillimanite zone. The leucogranite (Imwon leucogranite) is peraluminous granite which has high alumina index (A/CNK=1.31-1.93) and positive discriminant factor value (DF > 0). Thus, leucogranite is S-type granite generated from metasedimentary rocks. Major and trace element diagram ($R_1-R_2$ diagram and Rb vs. Y+Nb etc.) show collisional environment such as syn-collisional or volcanic arc granite. Because Rb/sr ratio (1.8-22.9) of leucogranites is higher than Sr/Ba ratio (0.21-0.79), leucogranite would be derived from muscovite dehydrate melting in metasedimentary rocks. Leucogranites have lower concentration of LREE and Eu and similar that of HREE relative to metasedimentary rocks. To examine difference of REEs between leucogranites and metasedimentary rocks, we perform modeling using volume percentage of a leucogranite and a metasedimenatry rock from study area and REE data of minerals from rhyolite (Nash and Crecraft, 1985) and melanosome of migmatite (Bea et al., 1994). Resultants of modeling indicate that LREE and HREE are controlled by monazites and garnet, respectively, although zircon is estimated HREE dominant in some leucogranite without garnet. Because there are many inclusions of accessary phases such as monazite and zircon in biotites from metasedimentary rocks. leucogranitic magma was mainly derived from muscovite-breakdown in metasedimenary rocks. Leucogranites can be subdivided into two types in compliance with Eu anomaly of chondrite nomalized REE pattern; the one of negative Eu anomaly is type I and the other is type II. Leucogranites have lower Eu concetnrations than that of metasedimenary rocks and similar that of both type. REE modeling suggest that this difference of Eu value is due to that of components of feldspars in both leucogranite and metasedimentary rock. The tendency of major ($K_2O$ and $Na_2O$) and face elements (Eu, Rb, Sr and Ba) of leucogranites also indicate that source magma of these two types was developed by anatexis experienced strong fractionation of alkali-feldspar. Conclusionally, leucogranites in this area are products of melts which was generated by muscovite-breakdown of metasedimenary rock in environment of continetal collision during high temperature/pressure metamorphism and then was fractionated and crystallized after extraction from source rock.

• Journal of Internet Computing and Services
• /
• v.16 no.6
• /
• pp.11-21
• /
• 2015

As growing interest in Human-Computer Interaction(HCI), research on HCI has been actively conducted. Also with that, research on Natural User Interface/Natural User eXperience(NUI/NUX) that uses user's gesture and voice has been actively conducted. In case of NUI/NUX, it needs recognition algorithm such as gesture recognition or voice recognition. However these recognition algorithms have weakness because their implementation is complex and a lot of time are needed in training because they have to go through steps including preprocessing, normalization, feature extraction. Recently, Kinect is launched by Microsoft as NUI/NUX development tool which attracts people's attention, and studies using Kinect has been conducted. The authors of this paper implemented hand-mouse interface with outstanding intuitiveness using the physical features of a user in a previous study. However, there are weaknesses such as unnatural movement of mouse and low accuracy of mouse functions. In this study, we designed and implemented a hand mouse interface which introduce a new concept called 'Virtual monitor' extracting user's physical features through Kinect in real-time. Virtual monitor means virtual space that can be controlled by hand mouse. It is possible that the coordinate on virtual monitor is accurately mapped onto the coordinate on real monitor. Hand-mouse interface based on virtual monitor concept maintains outstanding intuitiveness that is strength of the previous study and enhance accuracy of mouse functions. Further, we increased accuracy of the interface by recognizing user's unnecessary actions using his concentration indicator from his encephalogram(EEG) data. In order to evaluate intuitiveness and accuracy of the interface, we experimented it for 50 people from 10s to 50s. As the result of intuitiveness experiment, 84% of subjects learned how to use it within 1 minute. Also, as the result of accuracy experiment, accuracy of mouse functions (drag(80.4%), click(80%), double-click(76.7%)) is shown. The intuitiveness and accuracy of the proposed hand-mouse interface is checked through experiment, this is expected to be a good example of the interface for controlling the system by hand in the future.

• Journal of Intelligence and Information Systems
• /
• v.27 no.3
• /
• pp.175-197
• /
• 2021

Recently, with the development of deep learning technology, research on unstructured data analysis is being actively conducted, and it is showing remarkable results in various fields such as classification, summary, and generation. Among various text analysis fields, text classification is the most widely used technology in academia and industry. Text classification includes binary class classification with one label among two classes, multi-class classification with one label among several classes, and multi-label classification with multiple labels among several classes. In particular, multi-label classification requires a different training method from binary class classification and multi-class classification because of the characteristic of having multiple labels. In addition, since the number of labels to be predicted increases as the number of labels and classes increases, there is a limitation in that performance improvement is difficult due to an increase in prediction difficulty. To overcome these limitations, (i) compressing the initially given high-dimensional label space into a low-dimensional latent label space, (ii) after performing training to predict the compressed label, (iii) restoring the predicted label to the high-dimensional original label space, research on label embedding is being actively conducted. Typical label embedding techniques include Principal Label Space Transformation (PLST), Multi-Label Classification via Boolean Matrix Decomposition (MLC-BMaD), and Bayesian Multi-Label Compressed Sensing (BML-CS). However, since these techniques consider only the linear relationship between labels or compress the labels by random transformation, it is difficult to understand the non-linear relationship between labels, so there is a limitation in that it is not possible to create a latent label space sufficiently containing the information of the original label. Recently, there have been increasing attempts to improve performance by applying deep learning technology to label embedding. Label embedding using an autoencoder, a deep learning model that is effective for data compression and restoration, is representative. However, the traditional autoencoder-based label embedding has a limitation in that a large amount of information loss occurs when compressing a high-dimensional label space having a myriad of classes into a low-dimensional latent label space. This can be found in the gradient loss problem that occurs in the backpropagation process of learning. To solve this problem, skip connection was devised, and by adding the input of the layer to the output to prevent gradient loss during backpropagation, efficient learning is possible even when the layer is deep. Skip connection is mainly used for image feature extraction in convolutional neural networks, but studies using skip connection in autoencoder or label embedding process are still lacking. Therefore, in this study, we propose an autoencoder-based label embedding methodology in which skip connections are added to each of the encoder and decoder to form a low-dimensional latent label space that reflects the information of the high-dimensional label space well. In addition, the proposed methodology was applied to actual paper keywords to derive the high-dimensional keyword label space and the low-dimensional latent label space. Using this, we conducted an experiment to predict the compressed keyword vector existing in the latent label space from the paper abstract and to evaluate the multi-label classification by restoring the predicted keyword vector back to the original label space. As a result, the accuracy, precision, recall, and F1 score used as performance indicators showed far superior performance in multi-label classification based on the proposed methodology compared to traditional multi-label classification methods. This can be seen that the low-dimensional latent label space derived through the proposed methodology well reflected the information of the high-dimensional label space, which ultimately led to the improvement of the performance of the multi-label classification itself. In addition, the utility of the proposed methodology was identified by comparing the performance of the proposed methodology according to the domain characteristics and the number of dimensions of the latent label space.