• Journal of the Korean Data and Information Science Society
• /
• v.27 no.3
• /
• pp.827-833
• /
• 2016

In this paper, we propose a deep least squares support vector machine (LS-SVM) for regression problems, which consists of the input layer and the hidden layer. In the hidden layer, LS-SVMs are trained with the original input variables and the perturbed responses. For the final output, the main LS-SVM is trained with the outputs from LS-SVMs of the hidden layer as input variables and the original responses. In contrast to the multilayer neural network (MNN), LS-SVMs in the deep LS-SVM are trained to minimize the penalized objective function. Thus, the learning dynamics of the deep LS-SVM are entirely different from MNN in which all weights and biases are trained to minimize one final error function. When compared to MNN approaches, the deep LS-SVM does not make use of any combination weights, but trains all LS-SVMs in the architecture. Experimental results from real datasets illustrate that the deep LS-SVM significantly outperforms state of the art machine learning methods on regression problems.

• Science of Emotion and Sensibility
• /
• v.20 no.2
• /
• pp.161-170
• /
• 2017

Emotion affects many parts of human life such as learning ability, behavior and judgment. It is important to understand human nature. Emotion can only be inferred from facial expressions or gestures, what it actually is. In particular, emotion is difficult to classify not only because individuals feel differently about emotion but also because visually induced emotion does not sustain during whole testing period. To solve the problem, we acquired bio-signals and extracted features from those signals, which offer objective information about emotion stimulus. The emotion pattern classifier was composed of unsupervised learning algorithm with hidden nodes and feature vectors. Restricted Boltzmann machine (RBM) based on probability estimation was used in the unsupervised learning and maps emotion features to transformed dimensions. The emotion was characterized by non-linear classifiers with hidden nodes of a multi layer neural network, named deep belief network (DBN). The accuracy of DBN (about 94 %) was better than that of back-propagation neural network (about 40 %). The DBN showed good performance as the emotion pattern classifier.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.18 no.3
• /
• pp.106-118
• /
• 2019

In recent years, research and development on image object recognition using artificial intelligence have been actively carried out, and it is expected to be used for road maintenance. Among them, artificial intelligence models for object detection of road surface are continuously introduced. In order to develop such object recognition algorithms, a backbone network that extracts feature maps is essential. In this paper, we will discuss how to select the appropriate neural network. To accomplish it, we compared with 4 different deep neural networks using 6,000 road surface damage images. Based on three evaluation methods for analyzing characteristics of neural networks, we propose a method to determine optimal neural networks. In addition, we improved the performance through optimal tuning of hyper-parameters, and finally developed a light backbone network that can achieve 85.9% accuracy of road surface damage classification.

• The Journal of Korean Institute of Information Technology
• /
• v.16 no.11
• /
• pp.11-21
• /
• 2018

Research is carried out to categorize voices using Deep Learning technology. The study examines neural network-based sound classification studies and suggests improved neural networks for voice classification. Related studies studied urban data classification. However, related studies showed poor performance in shallow neural network. Therefore, in this paper the first preprocess voice data and extract feature value. Next, Categorize the voice by entering the feature value into previous sound classification network and proposed neural network. Finally, compare and evaluate classification performance of the two neural networks. The neural network of this paper is organized deeper and wider so that learning is better done. Performance results showed that 84.8 percent of related studies neural networks and 91.4 percent of the proposed neural networks. The proposed neural network was about 6 percent high.

• Smart Media Journal
• /
• v.7 no.1
• /
• pp.16-23
• /
• 2018

In this paper, we used the Deep Neural Network (DNN) to predict the number of daily spectators of Gwangju - KIA Champions Field in order to provide marketing data for the team and related businesses and for managing the inventories of the facilities in the stadium. In this study, the DNN model, which is based on an artificial neural network (ANN), was used, and four kinds of DNN model were designed along with dropout and batch normalization model to prevent overfitting. Each of four models consists of 10 DNNs, and we added extra models with ensemble model. Each model was evaluated by Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE). The learning data from the model randomly selected 80% of the collected data from 2008 to 2017, and the other 20% were used as test data. With the result of 100 data selection, model configuration, and learning and prediction, we concluded that the predictive power of the DNN model with ensemble model is the best, and RMSE and MAPE are 15.17% and 14.34% higher, correspondingly, than the prediction value of the multiple linear regression model.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2019.05a
• /
• pp.156-156
• /
• 2019

In this study, a gated recurrent unit (GRU) network is constructed based on a deep neural network (DNN) with the aim of restoring the missing daily flow data in river basins. Lai Chau hydrological station is located upstream of the Da river basin (Vietnam) is selected as the target station for this study. Input data of the model are data on observed daily flow for 24 years from 1961 to 1984 (before Hoa Binh dam was built) at 5 hydrological stations, in which 4 gauge stations in the basin downstream and restoring - target station (Lai Chau). The total available data is divided into sections for different purposes. The data set of 23 years (1961-1983) was employed for training and validation purposes, with corresponding rates of 80% for training and 20% for validation respectively. Another data set of one year (1984) was used for the testing purpose to objectively verify the performance and accuracy of the model. Though only a modest amount of input data is required and furthermore the Lai Chau hydrological station is located upstream of the Da River, the calculated results based on the suggested model are in satisfactory agreement with observed data, the Nash - Sutcliffe efficiency (NSE) is higher than 95%. The finding of this study illustrated the outstanding performance of the GRU network model in recovering the missing flow data at Lai Chau station. As a result, DNN models, as well as GRU network models, have great potential for application within the field of hydrology and hydraulics.

• Smart Structures and Systems
• /
• v.32 no.3
• /
• pp.179-193
• /
• 2023

This study explores an alternative to the existing centralized process for data anomaly detection in modern Internet of Things (IoT)-based structural health monitoring (SHM) systems. An edge intelligence framework is proposed for the early detection and classification of various data anomalies facilitating quality enhancement of acquired data before transmitting to a central system. State-of-the-art deep neural network pruning techniques are investigated and compared aiming to significantly reduce the network size so that it can run efficiently on resource-constrained edge devices such as wireless smart sensors. Further, depthwise separable convolution (DSC) is invoked, the integration of which with advanced structural pruning methods exhibited superior compression capability. Last but not least, quantization-aware training (QAT) is adopted for faster processing and lower memory and power consumption. The proposed edge intelligence framework will eventually lead to reduced network overload and latency. This will enable intelligent self-adaptation strategies to be employed to timely deal with a faulty sensor, minimizing the wasteful use of power, memory, and other resources in wireless smart sensors, increasing efficiency, and reducing maintenance costs for modern smart SHM systems. This study presents a theoretical foundation for the proposed framework, the validation of which through actual field trials is a scope for future work.

• International Journal of Computer Science & Network Security
• /
• v.23 no.8
• /
• pp.113-120
• /
• 2023

One of the most prevalent disease among women that leads to death is breast cancer. It can be diagnosed by classifying tumors. There are two different types of tumors i.e: malignant and benign tumors. Physicians need a reliable diagnosis procedure to distinguish between these tumors. However, generally it is very difficult to distinguish tumors even by the experts. Thus, automation of diagnostic system is needed for diagnosing tumors. This paper attempts to improve the accuracy of breast cancer detection by utilizing deep learning convolutional neural network (CNN). Experiments are conducted using Wisconsin Diagnostic Breast Cancer (WDBC) dataset. Compared to existing techniques, the used of CNN shows a better result and achieves 99.66%% in term of accuracy.

• The Journal of the Acoustical Society of Korea
• /
• v.39 no.6
• /
• pp.505-514
• /
• 2020

Since a large amount of training data is typically needed to train Deep Neural Networks (DNNs), a parallel training approach is required to train the DNNs. The Stochastic Gradient Descent (SGD) algorithm is one of the most widely used methods to train the DNNs. However, since the SGD is an inherently sequential process, it requires some sort of approximation schemes to parallelize the SGD algorithm. In this paper, we review various efforts on parallelizing the SGD algorithm, and analyze the computational overhead, communication overhead, and the effects of the approximations.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.26 no.1
• /
• pp.197-208
• /
• 2016

In this paper, we propose a method based on a convolutional neural network which is one of the deep neural network. So, we convert a malware code to malware image and train the convolutional neural network. In experiment with classify 9-families, the proposed method records a 96.2%, 98.7% of top-1, 2 error rate. And our model can classify 27 families with 82.9%, 89% of top-1,2 error rate.