• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.16 no.3
• /
• pp.229-236
• /
• 2003

Artificial neural network has been used for damage assessment by many researchers, but there are still some barriers that must be overcome to improve its accuracy and efficiency. The major problems associated with the conventional artificial neural network, especially the Back Propagation Neural Network(BPNN), are on the need of many training patterns and on the ambiguous relationship between neural network architecture and the convergence of solution. Therefore, the number of hidden layers and nodes in one hidden layer would be determined by trial and error. Also, it takes a lot of time to prepare many training patterns and to determine the optimum architecture of neural network. To overcome these drawbacks, the PNN can be used as a pattern classifier. In this paper, the PNN is used numerically to detect damage in a plate girder railway bridge. Also, the comparison between mode shapes and natural frequencies of the structure is investigated to select the appropriate training pattern for the damage detection in the railway bridge.

• Nuclear Engineering and Technology
• /
• v.51 no.3
• /
• pp.807-817
• /
• 2019

Airborne Gamma Ray Spectrometry (AGRS) with its important applications such as gathering radiation information of ground surface, geochemistry measuring of the abundance of Potassium, Thorium and Uranium in outer earth layer, environmental and nuclear site surveillance has a key role in the field of nuclear science and human life. The Broyden-Fletcher-Goldfarb-Shanno (BFGS), with its advanced numerical unconstrained nonlinear optimization in collaboration with Artificial Neural Networks (ANNs) provides a noteworthy opportunity for modern AGRS. In this study a new AGRS system empowered by ANN-BFGS has been proposed and evaluated on available empirical AGRS data. To that effect different architectures of adaptive ANN-BFGS were implemented for a sort of published experimental AGRS outputs. The selected approach among of various training methods, with its low iteration cost and nondiagonal scaling allocation is a new powerful algorithm for AGRS data due to its inherent stochastic properties. Experiments were performed by different architectures and trainings, the selected scheme achieved the smallest number of epochs, the minimum Mean Square Error (MSE) and the maximum performance in compare with different types of optimization strategies and algorithms. The proposed method is capable to be implemented on a cost effective and minimum electronic equipment to present its real-time process, which will let it to be used on board a light Unmanned Aerial Vehicle (UAV). The advanced adaptation properties and models of neural network, the training of stochastic process and its implementation on DSP outstands an affordable, reliable and low cost AGRS design. The main outcome of the study shows this method increases the quality of curvature information of AGRS data while cost of the algorithm is reduced in each iteration so the proposed ANN-BFGS is a trustworthy appropriate model for Gamma-ray data reconstruction and analysis based on advanced novel artificial intelligence systems.

• The Journal of the Acoustical Society of Korea
• /
• v.38 no.4
• /
• pp.433-441
• /
• 2019

In this paper, a method of estimating speech parameters for ultrasonic Doppler signals reflected from the articulatory muscles using LSTM (Long Short Term Memory) RNN (Recurrent Neural Networks) was introduced and compared with the method using MLP (Multi-Layer Perceptrons). LSTM RNN were used to estimate the Fourier transform coefficients of speech signals from the ultrasonic Doppler signals. The log energy value of the Mel frequency band and the Fourier transform coefficients, which were extracted respectively from the ultrasonic Doppler signal and the speech signal, were used as the input and reference for training LSTM RNN. The performance of LSTM RNN and MLP was evaluated and compared by experiments using test data, and the RMSE (Root Mean Squared Error) was used as a measure. The RMSE of each experiment was 0.5810 and 0.7380, respectively. The difference was about 0.1570, so that it confirmed that the performance of the method using the LSTM RNN was better.

• KIPS Transactions on Software and Data Engineering
• /
• v.11 no.10
• /
• pp.427-436
• /
• 2022

Korea Trade-Investment Promotion Agency (KOTRA) annually publishes the trade data in South Korea under the guidance of the Ministry of Trade, Industry and Energy in South Korea. The trade data usually contains Gross domestic product (GDP), a custom tariff, business score, and the price of export items in previous and this year, with regards to the trading items and the countries. However, it is challenging to figure out the meaningful insight so as to predict the future price on trading items every year due to the significantly large amount of data accumulated over the several years under the limited human/computing resources. Within this context, this paper proposes a multi layer perception that can predict the future price of potential trading items in the next year by training large amounts of past year's data with a low computational and human cost.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.5
• /
• pp.1431-1445
• /
• 2022

We construct a deep neural network model named ECGResNet. This model can diagnosis diseases based on 12-lead ECG data of eight common cardiovascular diseases with a high accuracy. We chose the 16 Blocks of ResNet50 as the main body of the model and added the Squeeze-and-Excitation module to learn the data information between channels adaptively. We modified the first convolutional layer of ResNet50 which has a convolutional kernel of 7 to a superposition of convolutional kernels of 8 and 16 as our feature extraction method. This way allows the model to focus on the overall trend of the ECG signal while also noticing subtle changes. The model further improves the accuracy of cardiovascular and cerebrovascular disease classification by using a fully connected layer that integrates factors such as gender and age. The ECGResNet model adds Dropout layers to both the residual block and SE module of ResNet50, further avoiding the phenomenon of model overfitting. The model was eventually trained using a five-fold cross-validation and Flooding training method, with an accuracy of 95% on the test set and an F1-score of 0.841.We design a new deep neural network, innovate a multi-scale feature extraction method, and apply the SE module to extract features of ECG data.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.20 no.1
• /
• pp.7-14
• /
• 2010

In this paper, we introduce a hierarchical internet application traffic classification system based on SVM as an alternative overcoming the uppermost limit of the conventional methodology which is using the port number or payload information. After selecting an optimal attribute subset of the bidirectional traffic flow data collected from the campus, the proposed system classifies the internet application traffic hierarchically. The system is composed of three layers: the first layer quickly determines P2P traffic and non-P2P traffic using a SVM, the second layer classifies P2P traffics into file-sharing, messenger, and TV, based on three SVDDs. The third layer makes specific classification of the entire 16 application traffics. By classifying the internet application traffic finely or coarsely, the proposed system can guarantee an efficient system resource management, a stable network environment, a seamless bandwidth, and an appropriate QoS. Also, even a new application traffic is added, it is possible to have a system incremental updating and scalability by training only a new SVDD without retraining the whole system. We validate the performance of our approach with computer experiments.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.11 no.3
• /
• pp.215-222
• /
• 2001

Neural network based models of semiconductor manufacturing processes have been shown to offer advantages in both accuracy and generalization over traditional methods. However, model development is often complicated by the fact that back-propagation neural networks contain several adjustable parameters whose optimal values unknown during training. These include learning rate, momentum, training tolerance, and the number of hidden layer neurOnS. This paper presents an investigation of the use of genetic algorithms (GAs) to determine the optimal neural network parameters for the modeling of plasma-enhanced chemical vapor deposition (PECVD) of silicon dioxide films. To find an optimal parameter set for the neural network PECVD models, a performance index was defined and used in the GA objective function. This index was designed to account for network prediction error as well as training error, with a higher emphasis on reducing prediction error. The results of the genetic search were compared with the results of a similar search using the simplex algorithm.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.39 no.3
• /
• pp.157-165
• /
• 2021

Recently, as the spatial resolution of satellite and aerial images has improved, various studies using remotely sensed data with high spatial resolution have been conducted. In particular, since the building extraction is essential for creating digital thematic maps, high accuracy of building extraction result is required. In this manuscript, building extraction models were generated using SegNet, U-Net, FC-DenseNet, and HRNetV2, which are representative semantic segmentation models in deep learning techniques, and then the evaluation of building extraction results was performed. Training dataset for building extraction were generated by using aerial orthophotos including various buildings, and evaluation was conducted in three areas. First, the model performance was evaluated through the region adjacent to the training dataset. In addition, the applicability of the model was evaluated through the region different from the training dataset. As a result, the f1-score of HRNetV2 represented the best values in terms of model performance and applicability. Through this study, the possibility of creating and modifying the building layer in the digital map was confirmed.

• The Journal of Korea Robotics Society
• /
• v.13 no.1
• /
• pp.16-25
• /
• 2018

The talking head (TH) indicates an utterance face animation generated based on text and voice input. In this paper, we propose the generation method of TH with facial expression and intonation by speech input only. The problem of generating TH from speech can be regarded as a regression problem from the acoustic feature sequence to the facial code sequence which is a low dimensional vector representation that can efficiently encode and decode a face image. This regression was modeled by bidirectional RNN and trained by using SAVEE database of the front utterance face animation database as training data. The proposed method is able to generate TH with facial expression and intonation TH by using acoustic features such as MFCC, dynamic elements of MFCC, energy, and F0. According to the experiments, the configuration of the BLSTM layer of the first and second layers of bidirectional RNN was able to predict the face code best. For the evaluation, a questionnaire survey was conducted for 62 persons who watched TH animations, generated by the proposed method and the previous method. As a result, 77% of the respondents answered that the proposed method generated TH, which matches well with the speech.

• Journal of Korean Tunnelling and Underground Space Association
• /
• v.19 no.1
• /
• pp.95-107
• /
• 2017

In this study, a preliminary study was undertaken for development of a tunnel incident automatic detection system based on a machine learning algorithm which is to detect a number of incidents taking place in tunnel in real time and also to be able to identify the type of incident. Two road sites where CCTVs are operating have been selected and a part of CCTV images are treated to produce sets of training data. The data sets are composed of position and time information of moving objects on CCTV screen which are extracted by initially detecting and tracking of incoming objects into CCTV screen by using a conventional image processing technique available in this study. And the data sets are matched with 6 categories of events such as lane change, stoping, etc which are also involved in the training data sets. The training data are learnt by a resilience neural network where two hidden layers are applied and 9 architectural models are set up for parametric studies, from which the architectural model, 300(first hidden layer)-150(second hidden layer) is found to be optimum in highest accuracy with respect to training data as well as testing data not used for training. From this study, it was shown that the highly variable and complex traffic and incident features could be well identified without any definition of feature regulation by using a concept of machine learning. In addition, detection capability and accuracy of the machine learning based system will be automatically enhanced as much as big data of CCTV images in tunnel becomes rich.