• Smart Structures and Systems
• /
• v.29 no.1
• /
• pp.181-193
• /
• 2022

Data-driven structural health monitoring (SHM) of civil infrastructure can be used to continuously assess the state of a structure, allowing preemptive safety measures to be carried out. Long-term monitoring of large-scale civil infrastructure often involves data-collection using a network of numerous sensors of various types. Malfunctioning sensors in the network are common, which can disrupt the condition assessment and even lead to false-negative indications of damage. The overwhelming size of the data collected renders manual approaches to ensure data quality intractable. The task of detecting and classifying an anomaly in the raw data is non-trivial. We propose an approach to automate this task, improving upon the previously developed technique of image-based pre-processing on one-dimensional (1D) data by enriching the features of the neural network input data with multiple channels. In particular, feature engineering is employed to convert the measured time histories into a 3-channel image comprised of (i) the time history, (ii) the spectrogram, and (iii) the probability density function representation of the signal. To demonstrate this approach, a CNN model is designed and trained on a dataset consisting of acceleration records of sensors installed on a long-span bridge, with the goal of fault detection and classification. The effect of imbalance in anomaly patterns observed is studied to better account for unseen test cases. The proposed framework achieves high overall accuracy and recall even when tested on an unseen dataset that is much larger than the samples used for training, offering a viable solution for implementation on full-scale structures where limited labeled-training data is available.

• ETRI Journal
• /
• v.31 no.3
• /
• pp.330-332
• /
• 2009

Broadcasting is one of the most important communication primitives used in multiprocessor networks. In this letter, we demonstrate that the broadcasting algorithm proposed by Madabhushi and others is incorrect. We introduce efficient one-to-all broadcasting schemes of even networks for one-port and all-port models. The broadcasting time of the one-port model is 2d-3 and that of the all-port model is d-1. The total time steps taken by the proposed algorithms are optimal.

• Journal of the Korea Society of Computer and Information
• /
• v.22 no.4
• /
• pp.41-49
• /
• 2017

Word sense disambiguation(WSD) that determines the exact meaning of homonym which can be used in different meanings even in one form is very important to understand the semantical meaning of text document. Many recent researches on WSD have widely used NNLM(Neural Network Language Model) in which neural network is used to represent a document into vectors and to analyze its semantics. Among the previous WSD researches using NNLM, RNN(Recurrent Neural Network) model has better performance than other models because RNN model can reflect the occurrence order of words in addition to the word appearance information in a document. However, since RNN model uses only the forward order of word occurrences in a document, it is not able to reflect natural language's characteristics that later words can affect the meanings of the preceding words. In this paper, we propose a WSD scheme using Bidirectional RNN that can reflect not only the forward order but also the backward order of word occurrences in a document. From the experiments, the accuracy of the proposed model is higher than that of previous method using RNN. Hence, it is confirmed that bidirectional order information of word occurrences is useful for WSD in Korean language.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.7 no.6
• /
• pp.1304-1311
• /
• 2003

Target tracking in a real world situation is difficult problem because of continuous variations in images, huge amounts of data, and high processing speed demands. The problem becomes even harder in the case of sea background. This paper presents an initial study of neural network based method for target detection and tracking in cluttering environment. The approach uses a combination of differential motion analysis, Kohonen self-organizing network and region growing method. The network is capable of detecting the mass-centers of moving objects within one frame. The history of neurons positions in the sequential frames approximates the traces of the targets. The experiments done with the network in simulated environment showed promising results.

• The Journal of the Korea Contents Association
• /
• v.8 no.4
• /
• pp.71-78
• /
• 2008

We propose the scheme that supports continuously service when a user tries to contact new target network in broadband convergence network even though the network resource for the service is not enough. The proposed scheme transforms original service capacity into suitable QoS for the target network using the scalable service techniques in the oncoding/decoding and supports handover for the user. We analyzed the handover blocking probability of two schemes using queueing system to show the improvement of performance by the proposed scheme. Subsequently, we showed that the handover blocking probability in the proposed scheme is lower than that in the existing scheme.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.11 no.6
• /
• pp.81-86
• /
• 2011

In one of the most impacting schemes proposed to address the TCP throughput problem over network coding, the network coding layer sends an acknowledgement if an innovative linear combination is received, even when a new packet is not decoded. Although this scheme is very effective, its implementation requires a limit on the coding window size. This limitation causes low TCP throughput in the presence of packet reordering. We argue that a TCP variant detecting a packet loss relying only on timers is effective in dealing with the packet reordering problem in network coding environments as well. Also we propose a new network coding layer to support such a TCP variant. Simulation results for a 2-path environment show that our proposed scheme improves TCP throughput by 19%.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2001.06a
• /
• pp.141-144
• /
• 2001

In this paper, we proposed a neural network controller with a fuzzy compensator which compensate a output of neural network controller. Even if learn by neural network controller, it can occur a bad results from disturbance or load variations. So in order to adjust above case. we used the fuzzy compensator to get an expected results. And the weight of main neural network can be changed with the result of learning an inverse model neural network of plant, so a expected dynamic characteristics of plant can be got. In order to confirm a performance of the proposed controller, we implemented the controller using the DSP processor and applied in a hydraulic servo system. And then we observed an experimental results.

• Proceedings of the IEEK Conference
• /
• 2002.07c
• /
• pp.2075-2078
• /
• 2002

Internet became the most popular network in spite of its weakness in realtime multimedia service. Many experts believe that the Internet has the potential to become the main multimedia distribution network of the near future. Currently, it does not provide any (BoS guarantees and, even when it does, guaranteed quality delivery of video may turn out to be too expensive. Unavoidable packet losses and delay jitter caused by congestion in a best effort delivery environment require use of intelligent transport techniques for effective video delivery. According to market needs of better quality of service (QoS) fur realtime multimedia services over Internet, they have been standardizing RSVP, IntServ, and DiffServ This paper combines the benefits of QoS mechanisms such as RSVP/IntServ with scalable video encoding. We propose that more important bit stream is given more priority such that limited network resources are guaranteed far the stream. Various prioritizing approaches are proposed and compared to normal approach by using Network Simulator. The calculated QoS parameters such as packet loss rate are used to calculate degree of degradation in video quality. In this Paper, proposed methods can be implemented adaptively to Von protocol, such as H.323, SIP.

• The Journal of the Acoustical Society of Korea
• /
• v.35 no.2
• /
• pp.143-148
• /
• 2016

Deep neural network that effectively capture the characteristics of data has been widely used in various applications. However, the amount of sound database is often insufficient for learning the deep neural network properly, so resulting in overfitting problems. In this paper, we propose a transfer learning framework that can effectively train the deep neural network even with insufficient sound event data by employing rich speech or music data. A series of experimental results verify that proposed method performs significantly better than the baseline deep neural network that was trained only with small sound event data.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.38B no.3
• /
• pp.172-179
• /
• 2013

In this paper, we consider network-coded bi-directional relaying (NCBR) schemes over an asymmetric channel, in which bi-directional links have the different channel quality, as well as the asymmetric traffic load. In order to deal with asymmetric nature, two different types of NCBR schemes are considered: network coding after padding (NaP) and network coding after fragmentation (NaF). Even if NaP has been known as only a useful means of dealing with the asymmetry in traffic load up to now, our analysis shows that its gain can be significantly lost by the asymmetry in channel quality, under the given bit error performance constraint. Furthermore, it is shown that NaF always outperforms NaP, as well as traditional bi-directional relaying scheme.