• International Journal of Concrete Structures and Materials
• /
• 제2권2호
• /
• pp.91-98
• /
• 2008

Curved bridges are important components of a highway transportation network for connecting local roads and highways, but very few data have been collected in terms of their field performance. This paper presents two-years monitoring and system identification results of a curved concrete box-girder bridge, the West St. On-Ramp, under ambient traffic excitations. The authors permanently installed accelerometers on the bridge from the beginning of the bridge life. From the ambient vibration data sets collected over the two years, the element stiffness correction factors for the columns, the girder, and boundary springs were identified using the back-propagation neural network. The results showed that the element stiffness values were nearly 10% different from the initial design values. It was also observed that the traffic conditions heavily influence the dynamic characteristics of this curved bridge. Furthermore, a probability distribution model of the element stiffness was established for long-term monitoring and analysis of the bridge stiffness change.

• 한국멀티미디어학회논문지
• /
• 제25권11호
• /
• pp.1643-1652
• /
• 2022

Most of the existing machine reading research has used Recurrent Neural Network (RNN) and Convolutional Neural Network (CNN) algorithms as networks. Among them, RNN was slow in training, and Question Answering Network (QANet) was announced to improve training speed. QANet is a model composed of CNN and self-attention. CNN extracts semantic and syntactic information well from the local corpus, but there is a limit to extracting the corresponding information from the global corpus. Graph Convolutional Networks (GCN) extracts semantic and syntactic information relatively well from the global corpus. In this paper, to take advantage of this strength of GCN, we propose I-QANet, which changed the CNN of QANet to GCN. The proposed model performed 1.2 times faster than the baseline in the Stanford Question Answering Dataset (SQuAD) dataset and showed 0.2% higher performance in Exact Match (EM) and 0.7% higher in F1. Furthermore, in the Korean Question Answering Dataset (KorQuAD) dataset consisting only of Korean, the learning time was 1.1 times faster than the baseline, and the EM and F1 performance were also 0.9% and 0.7% higher, respectively.

• 한국군사과학기술학회지
• /
• 제27권2호
• /
• pp.256-264
• /
• 2024

In the last 10 years, AI has made rapid progress, and image classification, in particular, are showing excellent performance based on deep learning. Nevertheless, due to the nature of deep learning represented by a black box, it is difficult to actually use it in critical decision-making situations such as national defense, autonomous driving, medical care, and finance due to the lack of explainability of judgement results. In order to overcome these limitations, in this study, a model description algorithm capable of local interpretation was applied to the inception network-derived AI to analyze what grounds they made when classifying national defense data. Specifically, we conduct a comparative analysis of explainability based on confidence values by performing LIME analysis from the Inception v2_resnet model and verify the similarity between human interpretations and LIME explanations. Furthermore, by comparing the LIME explanation results through the Top1 output results for Inception v3, Inception v2_resnet, and Xception models, we confirm the feasibility of comparing the efficiency and availability of deep learning networks using XAI.

• Bulletin of the Korean Chemical Society
• /
• 제33권2호
• /
• pp.613-619
• /
• 2012

We applied several machine learning methods for developing QSAR models for prediction of acute toxicity to fathead minnow. The multiple linear regression (MLR) and artificial neural network (ANN) method were applied to predict 96 h $LC_{50}$ (median lethal concentration) of 555 chemical compounds. Molecular descriptors based on 2D chemical structure were calculated by PreADMET program. The recursive partitioning (RP) model was used for grouping of mode of actions as reactive or narcosis, followed by MLR method of chemicals within the same mode of action. The MLR, ANN, and two RP-MLR models possessed correlation coefficients ($R^2$) as 0.553, 0.618, 0.632, and 0.605 on test set, respectively. The consensus model of ANN and two RP-MLR models was used as the best model on training set and showed good predictivity ($R^2$=0.663) on the test set.

• 농촌계획
• /
• 제21권3호
• /
• pp.171-179
• /
• 2015

The objectives of this research have been focussed on 1) developing prediction techniques for the flash flood and landslide based on rainfall prediction data in agricultural area and 2) developing an integrated forecasting system for the abrupt disasters using USN based real-time disaster sensing techniques. This study contains following steps to achieve the objective; 1) selecting rainfall prediction data, 2) constructing prediction techniques for flash flood and landslide, 3) developing USN and communication network protocol for detecting the abrupt disaster suitable for rural area, & 4) developing mobile application and SMS based early warning service system for local resident and tourist. Local prediction model (LDAPS, UM1.5km) supported by Korean meteorological administration was used for the rainfall prediction by considering spatial and temporal resolution. NRCS TR-20 and infinite slope stability analysis model were used to predict flash flood and landslide. There are limitations in terms of communication distance and cost using Zigbee and CDMA which have been used for existing disaster sensors. Rural suitable sensor-network module for water level and tilting gauge and gateway based on proprietary RF network were developed by consideration of low-cost, low-power, and long-distance for communication suitable for rural condition. SMS & mobile application forecasting & alarming system for local resident and tourist was set up for minimizing damage on the critical regions for abrupt disaster. The developed H/W & S/W for integrated abrupt disaster forecasting & alarming system was verified by field application.

• 한국실내디자인학회논문집
• /
• 제26권3호
• /
• pp.45-54
• /
• 2017

A social phenomenon that occurs in a physical space is said to be a complex system. However, space syntax, which is commonly employed by researchers to identify such social phenomena, has various limitations in interpreting their complexity. On the other hand, agent-based modeling considers a variety of factors including the personality of the agent, objective-oriented work flows, estimation according to time flows and better prediction of space use through diverse parameters depending the situation, as well as the characteristics of the space. The agent-based method thus has the potentials to be developed as an alternative to space syntax techniques. In particular, discrete event driven simulation(DEVS), which is part of the agent-based modeling method, embraces the concept of networks just like space syntax, which allows a possible theoretical linkage in the future. This study suggests a procedural model of agent-based DEVS reflecting two different connection methods, i.e. connections between adjacent areas and those of the entire space, and attempts to identify the relationship between the local and regional indices of space syntax. A number of spaces were selected as examples-one for a preliminary experiment and eight modified for the main experiment-and space syntax and DEVS were applied to each of them. The comparative analysis of the results led to the conclusions as follows: 1) Adjacent connections were closely related to local indices, while the whole-space approach to regional indices. Local integration shows both characteristics. 2) Observation of the time flow model indicated a faster convergence with the range of 1 to 3-fold of the total time of one lap, with the error of less than 10%. 3) The heat map analysis showed more obvious characteristics of using the space for the entire space rather than adjacent connections. 4) Space syntax shows higher eligibility than ABM.

• 유통과학연구
• /
• 제14권8호
• /
• pp.35-44
• /
• 2016

Purpose - This study aims to understand the relationship between Chinese local SNS usage and social capital building through Chinese international students in South Korea. A research model that illustrates the relationship between the SNS usage (i.e., intensity, communication and social capital building is proposed. Based on the analysis, this study will provide responses to the question of if SNS really presents the danger of trapping international consumers in their local comfort zone or enhance social capital for the users. Research design, data, and methodology - The survey questionnaire is circulated among the WeChat (a Chinese local SNS) users who are the Chinese international students studying in South Korea. The collected data is analyzed by structural equation method using SPSS and AMOS. Results - Proposed hypotheses of the positive relationships between the attachment of SNS use and both individuals' bridging and bonding social capital are supported. It's also supported that (1) interpersonal communication, (2) interpersonal communication with old friends, and (3) interpersonal communication for making new friends on SNS positively influence individuals' bridging social capital. Conclusions - This paper demonstrates the importance of intensity of WeChat use and interpersonal communication that impact Chinese international students' bridging and bonding social capital on WeChat.

• 한국정보통신학회논문지
• /
• 제8권8호
• /
• pp.1802-1809
• /
• 2004

음성 및 데이터 비용분리는 이동망 착신접속료 산정에 있어 음성트래픽과 관련된 비용만을 접속원가에 포함시키기 위한 것이다. 이에 대한 방안으로는 단일통화량 기준, 설비기준, 수익기준, 램지기준, 초고속망 비용배분 벤치마킹 등 여러가지가 있을 수 있으며 각각의 장단점을 비교하는 것이 중요하다. 본 연구에서는 먼저 비용배분에 대한 이론적 고찰을 수행하고 각각의 장단점 분석을 수행하고 있다. 또한 해외 정책적 연구와 관련된 사례분석을 통해 해외에서 적용하고 있는 논리를 검토하고 국내에 있어 도입시 고려해야 할 요소에 대해 살펴보고 있다. 초기에는 영국의 망세분화시 공유회선의 제공에 대해 50:50의 공통비용 배분방식을 도입하는 것을 고려해 볼 만하다. 이후에는 음성과 데이터 트래픽이 같아지는 시점이 오면 비율배분방식을 적용함으로써 증가된 데이터 트래픽에 따른 비용발생을 적절하게 고려할 수 있을 것이다.

• 한국멀티미디어학회논문지
• /
• 제7권12호
• /
• pp.1719-1728
• /
• 2004

The fusion of Internet technology and applications with wireless communication provides a new business model and promises to extend the possibilities of commerce to what is popularly called mobile commerce, or m-commerce. In mobile Internet banking service through wireless local area network, security is a most important factor to consider. We describe the development of security service for mobile Internet banking on Personal Digital Assistants (PDAs). Banking Server and Authentication Server were developed to simulate banking business and to support certificate management of authorized clients, respectively. To increase security, we took hybrid approach in implementation: symmetric block encryption and public-key encryption. Hash function and random number generation were exploited to generate a secret key. The data regarding banking service were encrypted with symmetric block encryption, RC4, and the random number sequence was done with public-key encryption. PDAs communicate through IEEE 802.IIb wireless LAN (Local Area Network) to access banking service. Several banking services and graphic user interfaces, which emulatedthe services of real bank, were developed to verity the working of each security service in PDA, the Banking Server, and the Authentication Server.

• Ocean Systems Engineering
• /
• 제4권1호
• /
• pp.53-62
• /
• 2014

The detection of ship wakes image can demonstrate substantial information regarding on a ship, such as its tonnage, type, direction, and speed of movement. Consequently, the wake target recognition is a favorable way for ship identification. This paper proposes a Local Binary Pattern (LBP) approach to extract image features (wakes) for training an Improved Back Propagation Artificial Neural Network (IBPANN) to identify ship speed. This method is applied to sort and recognize the ship wakes of five different speeds images, the result shows that the detection accuracy is satisfied as expected, the average correctness rates of wakes target recognition at the five speeds may be achieved over 80%. Specifically, the lower ship's speed, the better accurate rate, sometimes it's accuracy could be close to 100%. In addition, one significant feature of this method is that it can receive a higher recognition rate than the nearest neighbor classification method.