• Proceedings of the Korean Society of Computer Information Conference
• /
• 2021.07a
• /
• pp.1-2
• /
• 2021

3D NAND 플래시 메모리는 플래너(2D) NAND 셀을 적층하는 방식으로 단위 면적당 고용량을 제공한다. 하지만 적층 공정의 특성상 각 레이어별 또는 셀 위치에 따라 오류 발생 빈도가 달라질 수 있는 문제가 있다. 이와 같은 현상은 플래시 메모리의 쓰기/지우기(P/E) 횟수가 증가할 수록 두드러진다. SSD와 같은 대부분의 플래시 기반 저장장치는 오류 교정을 위하여 ECC를 사용한다. 이 방법은 모든 플래시 메모리 페이지에 대하여 고정된 보호 강도를 제공하므로 물리적 위치에 따라 에러 발생률이 각기 다르게 나타나는 3D NAND 플래시 메모리에서는 한계를 보인다. 따라서 본 논문에서는 오류 발생률 차이를 보이는 페이지와 레이어를 분류하여 각 영역별로 차별화된 보호강도를 적용한다. 우리는 페이지와 레이어별로 오류 발생률이 현저하게 달라지는 3K P/E 사이클에서 측정된 오류율을 바탕으로 페이지와 레이어를 분류하고 오류에 취약한 영역에 대해서는 패리티 데이터를 추가하여 차별화된 보호 강도를 제공한다. 오류 발생 횟수에 따른 영역 구분을 위하여 K-Means 머신러닝 알고리즘을 사용한다. 우리는 이와 같은 차별화된 보호정책이 3D NAND 플래시 메모리의 신뢰성과 수명향상에 기여할 수 있는 가능성을 보인다.

• The Journal of the Convergence on Culture Technology
• /
• v.10 no.2
• /
• pp.505-510
• /
• 2024

Since rolling contact fatigue cracks can always occur on the rail surface, which is the contact surface between wheels and rails, railway rails require thorough inspection and diagnosis to thoroughly inspect the condition of the cracks and prevent breakage. Recent detailed guidelines on the performance evaluation of track facilities present the requirements for methods and procedures for track performance evaluation. However, diagnosing and grading rail surface damage mainly relies on external inspection (visual inspection), which inevitably relies on qualitative evaluation based on the subjective judgment of the inspector. Therefore, in this study, we conducted a deep learning model study for rail surface defect detection using Fast R-CNN. After building a dataset of rail surface defect images, the model was tested. The performance evaluation results of the deep learning model showed that mAP was 94.9%. Because Fast R-CNN has a high crack detection effect, it is believed that using this model can efficiently identify rail surface defects.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2023.05a
• /
• pp.455-455
• /
• 2023

본 연구에서는 도시홍수 피해 저감 및 회복을 위한 도시홍수 연관 데이터 가시화 및 GIS 기반 LoD 1 수준 가시화 기술 개발을 진행하였다. 도시홍수는 불투수지역의 증가로 인한 첨두 홍수의 증가 및 도달 시간의 단축, 도시 내수배제의 불량으로 인한 주택지 및 상가 공장지 등의 침수에 의한 피해가 발생하는 현상이며, 도시홍수 예측 모델을 수행하기 위하여 수집한 기상, 하천 및 수자원, 토양 등의 데이터를 2차원 가시화하고 도심 지역의 지형 DEM(Digital Elevation Model) 데이터 및 건축물 DSM(Digital Surface Model) 데이터를 기반으로 3D 가시화를 진행하였다. 기상, 하천 및 수자원 관측 등의 데이터를 실시간으로 수집하며 관련 데이터를 도시홍수 디지털 트윈 플랫폼의 수문기상정보를 통하여 가시화 제공하며 토양 및 지리정보는 WMS 레이어를 기반으로 2D 가시화한다. 건축물 데이터의 경우 GIS 정보를 기반으로 하는 3D 객체 배치를 위하여 WGS84 타원체를 활용하여 EPSG:4326 좌표계를 적용하여 가시화하였다. 건축물 가시화는 PostgreSQL로 구축된 데이터를 Geoserver를 활용하여 자동으로 층 정보를 통한 건축물의 높이를 계산하도록 하였으며, CesiumJS를 적용하여 웹 기반 도시홍수 디지털 트윈 플랫폼을 개발하였고 추후 LoD 3 수준으로의 확대 적용 기반을 마련하였다.

• Korean Journal of Cognitive Science
• /
• v.28 no.4
• /
• pp.299-314
• /
• 2017

Predicting compound-protein interactions in-silico is significant for the drug discovery. In this paper, we propose an scalable machine learning model to predict compound-protein interaction. The key idea of this scalable machine learning model is the architecture of pairwise neural network model and feature embedding method from the raw data, especially for protein. This method automatically extracts the features without additional knowledge of compound and protein. Also, the pairwise architecture elevate the expressiveness and compact dimension of feature by preventing biased learning from occurring due to the dimension and type of features. Through the 5-fold cross validation results on large scale database show that pairwise neural network improves the performance of predicting compound-protein interaction compared to previous prediction models.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.1 no.1
• /
• pp.109-117
• /
• 1998

To obtain new information using a single remotely sensed image data is limited to extract various information. Recent trends in the remote sensing show that many researchers integrate and analyze many different forms of remotely sensed data, such as optical and radar satellite images, aerial photograph, airborne multispectral scanner data and land spectral scanners. Korean researchers have not been using such a combined dataset yet. This study intended to apply the technique of integration between optical data and radar data(SAR) and to examine the output that had been obtained through the technique of supervised classification using the result of integration. As a result, we found of better enhanced image classification results by using IHS conversion than by using RGB mixed and interband correlation.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.11 no.1
• /
• pp.56-65
• /
• 2007

Existing application systems of enterprises can process and collect a lot of information within enterprises, but it could be difficult to share validated information with partners and other system in the process of contacting to other legacy systems required in cooperation environment. For solving these problems, EAI systems are introduced in cooperation environment so that data sharing and integration can be achieved. The integration based on EAI is not limited to particular business system but all systems of an enterprise so that standard of suitable metadata level is needed for forwarding consistently between each business of systems. Therefore this paper maintains consistency of data sharing and integrating among legacy systems in cooperative environment for proposing message exchanging based on bridge XMDR.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.17 no.2
• /
• pp.95-106
• /
• 2014

In this study, the applicability of a standard vector model was evaluated using RIMGIS vector data and a portal based river-geospatial information web service system was developed using XML and JSON based data linkage between the server and the client. The RIMGIS vector data including points, lines, and polygons were converted to the Geospatial Data Model(GDM) developed in this study and were validated by layers. After the conversion, it was identified that the attribute data of a shape file remained without loss. The GeoServer GDB(GeoDataBase) that manages a DB in the portal was developed as a management module. The XML-based Geography Markup Language(GML) standards of OGC was used for accessing to and managing vector layers and encoding spatial data. The separation of data content and expression in the GML allowed the different expressions of the same data, convenient data revision and update, and enhancing the expandability. In the future, it is necessary to improve the access, exchange, and storage of river-geospatial information through the user's customized services and Internet accessibility.

• Journal of the Korean Institute of Gas
• /
• v.14 no.2
• /
• pp.40-46
• /
• 2010

In this study, the integrated control and safety management system for a super-large LNG membrane storage tank has been presented based on the investigation and analysis of measuring equipments and safety analysis system for a conventional LNG membrane storage tank. The integrated control and safety management system, which may increase a safety and efficiency of a super-large LNG membrane storage tank, added additional pressure gauges and new displacement/force sensors at the steel anchor between an inner tank and a prestressed concrete structure. The displacement and force sensors may provide clues of a membrane panel failure and a LNG leakage from the inner tank. The conventional leak sensor may not provide proper information on the membrane panel fracture even though LNG is leaked until the leak detector, which is placed at the insulation area behind an inner tank, send a warning signal. Thus, the new integrated control and safety management system is to collect and analyze the temperature, pressure, displacement, force and LNG density, which are related to the tank system safety and leakage control from the inner tank. The digital data are also measured from measurement systems such as displacement and force of a membrane panel safety, LNG level and density, cool-down process, leakage, and pressure controls.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.30 no.4
• /
• pp.379-388
• /
• 2012

The Road Name Address system has begun to be applied and widely used since 2011. However, the Digital Map, or the national representative basic map, has no reference to the Road Name Address system. It causes some difficulties to use the Digital Map under the Road Name Address system. In this paper, we suggest a method for generating the expanded Digital Map by adding information about Road Name Address system into the objects of the Digital Map. First, object matching pairs between the road section layer from the Road Name Address Map and the road centerline layer from the Digital Map are found. Then attributes to be copied from the Road Name Address map to the Digital Map are extracted by comparing their attribute tables. Finally the extracted attributes are copied from the Road Name Address Map to the Digital Map. The expanded road centerline layer of the Digital Map then has attributes about road name according to the Road Name Address system, so that the georeferencing of the Digital Map according to the Road Name Address system becomes possible.

• The Transactions of the Korea Information Processing Society
• /
• v.13 no.9
• /
• pp.444-452
• /
• 2024

This study proposes a deep learning architecture optimized for fire detection derived through Layer Importance Evaluation. In order to solve the problem of unnecessary complexity and operation of the existing Convolutional Neural Network (CNN)-based fire detection system, the operation of the inner layer of the model based on the weight and activation values was analyzed through the Layer Importance Evaluation technique, the layer with a high contribution to fire detection was identified, and the model was reconstructed only with the identified layer, and the performance indicators were compared and analyzed with the existing model. After learning the fire data using four transfer learning models: Xception, VGG19, ResNet, and EfficientNetB5, the Layer Importance Evaluation technique was applied to analyze the weight and activation value of each layer, and then a new model was constructed by selecting the top rank layers with the highest contribution. As a result of the study, it was confirmed that the implemented architecture maintains the same performance with parameters that are about 80% lighter than the existing model, and can contribute to increasing the efficiency of fire monitoring equipment by outputting the same performance in accuracy, loss, and confusion matrix indicators compared to conventional complex transfer learning models while having a learning speed of about 3 to 5 times faster.