• Journal of Sensor Science and Technology
• /
• v.30 no.5
• /
• pp.286-289
• /
• 2021

Image classification is one of the fundamental applications of computer vision. It enables a system to identify an object in an image. Recently, image classification applications have broadened their scope from computer applications to edge devices. The convolutional neural network (CNN) is the main class of deep learning neural networks that are widely used in computer tasks, and it delivers high accuracy. However, CNN algorithms use a large number of parameters and incur high computational costs, which hinder their implementation in edge hardware devices. To address this issue, this paper proposes a lightweight image classifier that provides good accuracy while using fewer parameters. The proposed image classifier diverts the input into three paths and utilizes different scales of receptive fields to extract more feature maps while using fewer parameters at the time of training. This results in the development of a model of small size. This model is tested on the CIFAR-10 dataset and achieves an accuracy of 90% using .26M parameters. This is better than the state-of-the-art models, and it can be implemented on edge devices.

• Structural Engineering and Mechanics
• /
• v.91 no.2
• /
• pp.163-175
• /
• 2024

The objective of this research is to improve public safety in civil engineering by recognizing fractures in concrete structures quickly and correctly. The study offers a new crack detection method based on advanced image processing and machine learning techniques, specifically transfer learning with convolutional neural networks (CNNs). Four pre-trained models (VGG16, AlexNet, ResNet18, and DenseNet161) were fine-tuned to detect fractures in concrete surfaces. These models constantly produced accuracy rates greater than 80%, showing their ability to automate fracture identification and potentially reduce structural failure costs. Furthermore, the study expands its scope beyond crack detection to identify concrete health, using a dataset with a wide range of surface defects and anomalies including cracks. Notably, using VGG16, which was chosen as the most effective network architecture from the first phase, the study achieves excellent accuracy in classifying concrete health, demonstrating the model's satisfactorily performance even in more complex scenarios.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.21 no.11
• /
• pp.2167-2175
• /
• 2017

In recent years, an IoT is dramatically developing according to the enhancement of AI, the increase of connected devices, and the high-performance cloud systems. Huge data produced by many devices and sensors is expanding the scope of services, such as an intelligent diagnostics, a recommendation service, as well as a smart monitoring service. The studies of edge computing are limited as a role of small server system with high quality HW resources. However, there are specialized requirements in a smart factory domain needed edge computing. The edges are needed to pre-process containing tiny filtering, pre-formatting, as well as merging of group contexts and manage the regional rules. So, in this paper, we extract the features and requirements in a scope of efficiency and robustness. Our edge offers to decrease a network resource consumption and update rules and learning models. Moreover, we propose architecture of edge computing based on learning factors sharing with a cloud system in a smart factory.

• Journal of Korea Port Economic Association
• /
• v.28 no.2
• /
• pp.113-128
• /
• 2012

By strengthening the market control and expanding the networks, providers of global logistics are expanding their service scope. Warehouses connects networks to between internal and external warehouses worldwide by using employees. The paper focuses on analyzing the efficiency of the top forty Global Logistics Providers. Therefore, the study classifies the factors which specify the efficiency of a total logistics industry and verified its firmness. Furthermore, the most recent published reports by Logistics Quarterly and Armstrong Association in 2012 was used in order to guarantee credibility of the study. For scope period for analysis, this study utilizes three years of materials, from 2007 to 2010. By applying the DEA Window model and Malmquist, the trend in efficiency and stableness was analyzed. Consequently, the main purpose of the paper is evaluating the efficiency. Also, analyzing its determinants and illustrating a long-term relationship between the annual turnover and the number of Industry Focus/Key Customers was used as output measures. In addition, the number of warehouses worldwide and employees worldwide, of Logistics Providers were used as input measures.

• Smart Structures and Systems
• /
• v.30 no.5
• /
• pp.545-556
• /
• 2022

Through the examination of literatures, electronic commerce is a subject which is accepted in enterprises to define e-commerce adoption, trends, and issues that are assisting and obstructing its efficacy. E-commerce offers numerous advantages to consumer satisfaction in any place and helps the company to get a competitive benefit over its competitors. The Internet has expanded the scope of business. Many business information is available by the global network that supports information gathering between organizations, businesses and their clients, while various divisions of a business is increasing at an exponential rate. Meanwhile, there are a few barriers to proper e-commerce usage and adoption, such as reliable internet connections, poor e-commerce supporting infrastructures, logistics systems presenting socio-regulatory and poor transportation barriers and demonstrating the significant improvement of e-commerce reliable and affordable Internet provisions, i.e., Internet cost, intensity, and reasonable level of e-readiness. The operational and strategic significance of information-based virtual value chains for all organizations cannot be emphasized. As a consequence, this study confirms worldwide market elements of e-commerce, such as its issues, benefits, relevance, scope, facilitators and projects prospective obstacles in a developing economy.

• Journal of KIISE:Information Networking
• /
• v.34 no.5
• /
• pp.405-422
• /
• 2007

Recently, according to the network environment, there are many researches for home network. Nowadays, in home network, the method that access control policy is managed for each home device by using ACL is popular, and EAM (Extranet access management) is applied as a solution. In addition, the research about secure OS is ongoing based on open operating system and the research of user authentication mechanisms for home network using home server is also in progress. However, these researches have some problems as follows; First, the transmission scope of expected access technology in home network is wide, so unauthenticated outside terminal can access the home network. Second, user is inconvenient because user need to set the necessary information for each device. Third, user privacy and convenience are not considered. OSGi provides a service platform for heterogeneous technologies in home network environment. Here, user access control is one of the core parts which should have no problems such as above items, but there are no concrete researches yet. Thus in this paper, we propose an access control policy management framework and access control operation based on RBAC for user access control in home network environment in which OSGi service platform is operated. First, we list the consideration which is not clearly mentioned in OSGi standard, and then we solve these above problems through new framework. In addition, we propose the effective and economical operation method which reduces the policy change frequency for user access control by using RBAC concept though limited resource of home gateway. Besides, in this paper, these proposed policies are defined separately as user-role assignment policy and permission-role assignment policy, and user decide their own policies. In conclusion, we provide the scheme to enhance the user convenience and to solve the privacy problem.

• Journal of the Korean Society for information Management
• /
• v.26 no.3
• /
• pp.295-316
• /
• 2009

This research applied the concepts of IT architecture and IT governance for managing with an integrated computing environment and organized structure, which base a digital library's management and operation. It also aims to analyze the structural system between information technology of human resources and strategy alignment elements of business, which both constitute the core content. Social network analysis software was used to investigate the complicated relationship between IT and business-related strategy alignment elements. The following is the results of carrying out this research on the social network structure and features of strategy alignment elements for a digital library. First, analysis indexes for strategy alignment elements and social network of a digital library were developed. Second, an analysis model was designed based on the analysis index for social network as to strategy alignment elements. Analysis model was appraised by collecting social network datasets for such strategy alignment elements as Communications, Competency and Value, Governance, Partnership, Scope and Architecture, and Skills against the Business strategy, Information strategy, Business and Technology of a digital library. As for the content of analysis, social network structure and specific features were analyzed in relation to a digital library's (1) General social network, (2) Structure of strategy alignment elements, (3) Strategy fit and Functional integration.

• Journal of Korean Society of Transportation
• /
• v.22 no.5
• /
• pp.111-122
• /
• 2004

The driver's decision making (e.g. route choice) is a typical decision making with an uncertainty. In this paper, we investigate the effect of route guidance information on driver's route choice and learning behavior and analyse the potential of information system in a road network in which traffic flows follow random walk. A Simulation performed focuses on the relationship among the network wide performance, message receiving rates and driver's learning mechanism. We know that at high levels of message receiving rates, the network-wide performance may get worse. However, at low levels of receiving rates, we found that the travel time when guidance information is provided decrease compared to the cases when no pubic information is provided. Also, we found that the learning parameter of the learning mechanism model always changes under nonstationary traffic condition. In addition, learning process of drivers does not converge on any specific value. More investigation is needed to enlarge the scope of the study and to explore more deeply driver's behavior.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.28 no.2
• /
• pp.457-469
• /
• 2018

This study aimed to investigate the fundamental reasons for the presence of multi-homed host and the risks associated with such risky system. Furthermore, multi-homed host detection methods that have been researched and developed so far were compared and analyzed to determine areas for improvement. Based on the results, we propose the model of an improved automatic detection system and we implemented it. The experimental environment was configured to simulate the actual network configuration and endpoints of an organization employing network segmentation. And the functionality and performance of the detection system were finally measured while generating multi-homed hosts by category, after the developed detection system had been installed in the experiment environment. We confirmed that the system work correctly without false-positive, false-negative in the scope of this study. To the best of our knowledge, the presented detection system is the first academic work targeting multi-homed host under agent-based.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2017.04a
• /
• pp.315-318
• /
• 2017

Nowadays, Distributed Denial of Service (DDoS) attacks have gained increasing popularity and have been a major factor in a number of massive cyber-attacks. It could easily exhaust the computing and communicating resources of a victim within a short period of time. Therefore, we have to find the method to detect and prevent the DDoS attack. Recently, there have been some researches that provide the methods to resolve above problem, but it still gets some limitations such as low performance of detecting and preventing, scope of method, most of them just use on cloud server instead of network, and the reliability in the network. In this paper, we propose solutions for (1) handling multiple DDoS attacks from multiple IP address and (2) handling the suspicious attacks in the network. For the first solution, we assume that there are multiple attacks from many sources at a times, it should be handled to avoid the conflict when we setup the preventing rule to switches. In the other, there are many attacks traffic with the low volume and same destination address. Although the traffic at each node is not much, the traffic at the destination is much more. So it is hard to detect that suspicious traffic with the sampling based method at each node, our method reroute the traffic to another server and make the analysis to check it deeply.