• Proceedings of the KIEE Conference
• /
• 2004.11c
• /
• pp.358-360
• /
• 2004

The digital images contain a significant amount of redundancy and require a large amount of data for their storage and transmission. Therefore, the image compression is necessary to treat digital images efficiently. The goal of image compression is to reduce the number of bits required for their representation. The image compression can reduce the size of image data using contractive mapping of original image. Among the compression methods, the mapping is affine transformation to find the block(called range block) which is the most similar to the original image. In this paper, we applied the neural network(SOM) in encoding. In order to improve the performance of image compression, we intend to reduce the similarities and unnecesaries comparing with the originals in the codebook. In standard image coding, the affine transform is performed with eight isometries that used to approximate domain blocks to range blocks.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.33 no.9C
• /
• pp.691-696
• /
• 2008

A new vocoder of Code Excited Linear Predictive (CELP) based on Adaptive Multi Rate (AMR) 7.4kbit/s mode is proposed in this paper. The proposed vocoder achieves a better compression rate in an environment of Speaker Dependent Coding System (SDSC) and is efficiently used for systems, such as OGM(Outgoing message) and TTS(Text To Speech), which needs only one person's speech. In order to enhance the compression rate of a coder, a new Line Spectral Pairs(LSP) code-book is employed by using Centroid Neural Network (CNN) algorithm. In comparison with original(traditional) AMR 7.4 Kbit/s coder, the new coder shows 27% higher compression rate while preserving synthesized speech quality in terms of Mean Opinion Score(MOS).

• Journal of the Korean Institute of Intelligent Systems
• /
• v.10 no.3
• /
• pp.179-187
• /
• 2000

Fractal theory has strengths such as high compression rate and fast decoding time in application to image compression, but it suffers from long comparison time necessary for finding an optimally similar domain block in the encoding stage. This paper proposes a neural network based block classifier which enhances the encoding time significantly by classifying domain blocks into 4 patterns and searching only those blocks having the same pattern with the range block to be encoded. Size of a block is differently determined depending on the image complexity of the block. The proposed algorithm has been tested with three different images having various featrues. The experimental results have shown that the proposed algorithm enhances the compression time by 40% on average compared to the conventional fractal encoding algorithms, while maintaining allowable image qualify of PSNR 30 dB.

• Computers and Concrete
• /
• v.22 no.6
• /
• pp.539-550
• /
• 2018

Concrete is highly non-linear material which is originating from the transition zone in the form of micro-cracks, governs material response under various loadings. In this paper, the constitutive models published by many researchers have been used to generate novel stiffness parameters and constitutive curves for concrete. Following such linear material formulations, where the energy is conservative during the curvature, and a nonlinear contribution to the concrete has been made and investigated. In which, nonlinear concrete elastic modulus modeling has been developed that is capable-of representing concrete elasticity for grades ranging from 10 to 140 MPa. Thus, covering the grades range of concrete up to the ultra-high strength concrete, and replacing many concrete models that are valid for narrow ranges of concrete strength grades. This has been followed by the introduction of the nonlinear Hooke's law for the concrete material through the replacement of the Young constant modulus with the nonlinear modulus. In addition, the concept of concrete elasticity index (${\varphi}$) has been proposed and this factor has been introduced to account for the degradation of concrete stiffness in compression under increased loading as well as the multi-stages micro-cracking behavior of concrete under uniaxial compression. Finally, a sub-routine artificial neural network model has been developed to capture the concrete behavior that has been introduced to facilitate the prediction of concrete properties under increased loading.

• Journal of the Korean Geotechnical Society
• /
• v.40 no.4
• /
• pp.127-135
• /
• 2024

The purpose of this paper is to establish the relationship between compression wave velocity and porosity in unsaturated soil using a deep neural network (DNN) algorithm. Input parameters were examined using the error norm method to assess their impact on porosity. Compression wave velocity was conclusively found to have the most significant influence on porosity estimation. These parameters were derived through both field and laboratory experiments using a total of 266 numerical data points. The application of the DNN was evaluated by calculating the mean squared error loss for each iteration, which converged to nearly zero in the initial stages. The predicted porosity was analyzed by splitting the data into training and validation sets. Compared with actual data, the coefficients of determination were exceptionally high at 0.97 and 0.98, respectively. This study introduces a methodology for predicting dependent variables through error norm analysis by disregarding fewer sensitive factors and focusing on those with greater influence.

• Steel and Composite Structures
• /
• v.46 no.2
• /
• pp.153-173
• /
• 2023

Rectangular concrete-filled steel tubular (RCFST) column, a type of concrete-filled steel tubular (CFST), is widely used in compression members of structures because of its advantages. This paper proposes a robust machine learning-based framework for predicting the ultimate compressive strength of RCFST columns under both concentric and eccentric loading. The gradient boosting neural network (GBNN), an efficient and up-to-date ML algorithm, is utilized for developing a predictive model in the proposed framework. A total of 890 experimental data of RCFST columns, which is categorized into two datasets of concentric and eccentric compression, is carefully collected to serve as training and testing purposes. The accuracy of the proposed model is demonstrated by comparing its performance with seven state-of-the-art machine learning methods including decision tree (DT), random forest (RF), support vector machines (SVM), deep learning (DL), adaptive boosting (AdaBoost), extreme gradient boosting (XGBoost), and categorical gradient boosting (CatBoost). Four available design codes, including the European (EC4), American concrete institute (ACI), American institute of steel construction (AISC), and Australian/New Zealand (AS/NZS) are refereed in another comparison. The results demonstrate that the proposed GBNN method is a robust and powerful approach to obtain the ultimate strength of RCFST columns.

• Transactions of Materials Processing
• /
• v.33 no.2
• /
• pp.96-102
• /
• 2024

This study addresses evaluation of performance of hardening model for a titanium alloy (Ti6Al4V) based on the artificial neural network (ANN) regarding the strain rate and the temperature. Uniaxial compression tests were carried out at different strain rates from 0.001 /s to 10 /s and temperatures from 575 ℃ To 975 ℃. Using the experimental data, ANN models were trained and tested with different hyperparameters, such as size of hidden layer and optimizer. The input features were determined with the equivalent plastic strain, strain rate, and temperature while the output value was set to the equivalent stress. When the number of data is sufficient with a smooth tendency, both the Bayesian regulation (BR) and the Levenberg-Marquardt (LM) show good performance to predict the flow behavior. However, only BR algorithm shows a predictability when the number of data is insufficient. Furthermore, a proper size of the hidden layer must be confirmed to describe the behavior with the limited number of the data.

• ETRI Journal
• /
• v.40 no.4
• /
• pp.421-434
• /
• 2018

The application of deep neural networks (DNNs) to connect the world with cyber physical systems (CPSs) has attracted much attention. However, DNNs require a large amount of memory and computational cost, which hinders their use in the relatively low-end smart devices that are widely used in CPSs. In this paper, we aim to determine whether DNNs can be efficiently deployed and operated in low-end smart devices. To do this, we develop a method to reduce the memory requirement of DNNs and increase the inference speed, while maintaining the performance (for example, accuracy) close to the original level. The parameters of DNNs are decomposed using a hybrid of canonical polyadic-singular value decomposition, approximated using a tensor power method, and fine-tuned by performing iterative one-shot hybrid fine-tuning to recover from a decreased accuracy. In this study, we evaluate our method on frequently used networks. We also present results from extensive experiments on the effects of several fine-tuning methods, the importance of iterative fine-tuning, and decomposition techniques. We demonstrate the effectiveness of the proposed method by deploying compressed networks in smartphones.

• Journal of Broadcast Engineering
• /
• v.25 no.2
• /
• pp.200-207
• /
• 2020

Recently, deep neural networks such as CNN are showing excellent performance in various fields such as image classification, object recognition, visual quality enhancement, etc. However, as the model size and computational complexity of deep learning models for most applications increases, it is hard to apply neural networks to IoT and mobile environments. Therefore, neural network compression algorithms for reducing the model size while keeping the performance have been being studied. In this paper, we apply few compression methods to CNN models and evaluate their performances in the embedded environment. For evaluate the performance, the classification performance and inference time of the original CNN models and the compressed CNN models on the image inputted by the camera are evaluated in the embedded board equipped with QCS605, which is a customized AI chip. In this paper, a few CNN models of MobileNetV2, ResNet50, and VGG-16 are compressed by applying the methods of pruning and matrix decomposition. The experimental results show that the compressed models give not only the model size reduction of 1.3~11.2 times at a classification performance loss of less than 2% compared to the original model, but also the inference time reduction of 1.2~2.21 times, and the memory reduction of 1.2~3.8 times in the embedded board.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• fall
• /
• pp.41-44
• /
• 2021

In this paper, we propose an efficient reference-based compression artifact reduction network for 360-degree images in an equi-rectangular projection (ERP) domain. In our insight, conventional image restoration methods cannot be applied straightforwardly to 360-degree images due to the spherical distortion. To address this problem, we propose an adaptive disparity estimator using a deformable convolution to exploit correlation among 360-degree images. With the help of the proposed convolution, the disparity estimator establishes the spatial correspondence successfully between the ERPs and extract matched textures to be used for image restoration. The experimental results demonstrate that the proposed algorithm provides reliable high-quality textures from the reference and improves the quality of the restored image as compared to the state-of-the-art single image restoration methods.