• Journal of KIISE:Software and Applications
• /
• v.31 no.10
• /
• pp.1386-1399
• /
• 2004

This paper presents correlation-based automatic image captioning. Given a training set of annotated images, we want to discover correlations between visual features and textual features, so that we can automatically generate descriptive textual features for a new unseen image. We develop models with multiple design alternatives such as 1) adaptively clustering visual features, 2) weighting visual features and textual features, and 3) reducing dimensionality for noise sup-Pression. We experiment thoroughly on 10 data sets of various content styles from the Corel image database, about 680MB. The major contributions of this work are: (a) we show that careful weighting visual and textual features, as well as clustering visual features adaptively leads to consistent performance improvements, and (b) our proposed methods achieve a relative improvement of up to 45% on annotation accuracy over the state-of-the-art, EM approach.

• The Korean Journal of Applied Statistics
• /
• v.32 no.6
• /
• pp.899-908
• /
• 2019

Massive in-depth neural networks with numerous parameters are powerful machine learning methods, but they have overfitting problems due to the excessive flexibility of the models. Dropout is one methods to overcome the problem of oversized neural networks. It is also an effective method that randomly drops input and hidden nodes from the neural network during training. Every sample is fed to a thinned network from an exponential number of different networks. In this study, instead of feeding one sample for each thinned network, two or more samples are used in fitting for one thinned network known as a Hybrid Dropout. Simulation results using real data show that the new method improves the stability of estimates and reduces the minimum error for the verification data.

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.55 no.4
• /
• pp.378-391
• /
• 2019

The objective of this study is to establish a direction for smart aquaculture technology development in the Republic of Korea through patent analysis and a demand survey of experts and fishermen. The patent analysis was conducted using Wisdomain for patents in the Republic of Korea, the United States of America, Europe, Japan, and China from 2005 to 2016. This study conducted an analytic hierarchy process (AHP) survey of experts in the fields of fishery, marine, and ICT among others. Furthermore, it carried out a demand survey of 85 fishermen in Jeonnam and Jeju. The smart aquaculture technology market has moderately grown in the Republic of Korea until recently, and it is expected to expand further because of the expansion of national investment in the smart aquaculture field. The priority evaluation results for developing smart aquaculture technology show that land-based aquaculture has a higher priority than sea-based aquaculture. Of the fishermen that responded, 84% said that they need to introduce smart aquaculture technology to solve problems in the supply and demand of manpower, labor cost, and maintenance expenses. The direction of development should lie in developing biological and environment-based standard aquaculture models to spread high-tech systems and vitalize the aquaculture industry. This requires continual training of human resources in the smart aquaculture field.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.48 no.8
• /
• pp.631-640
• /
• 2020

Currently, aircraft simulator has drawn a great attention because it has significant advantages of economic, temporal, and spatial costs compared with pilot training with real aircraft. Among the components of the aircraft simulator, flight dynamic model plays a key role in simulating the flight of an actual aircraft. Hence, it is important to verify the fidelity of flight dynamic model with an automated tool. In this paper, we develop a software to automatically verify the fidelity of the flight mechanics model for the efficient development of the aircraft simulator. After designing the software structure and GUI based on the requirements derived from the fidelity verification process, the software is implemented with C # language in Window-based environment. Experimental results on CTSW models show that the developed software is effective in terms of function, performance and user convenience.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.10 no.4
• /
• pp.1824-1845
• /
• 2016

The GMM is a conventional approach which has been recently applied in many face recognition studies. However, the question about how to deal with illumination changes while ensuring high performance is still a challenge, especially with real-world databases. In this paper, we propose a Visual Observation Confidence (VOC) measure for robust face recognition for illumination changes. Our VOC value is a combined confidence value of three measurements: Flatness Measure (FM), Centrality Measure (CM), and Illumination Normality Measure (IM). While FM measures the discrimination ability of one face, IM represents the degree of illumination impact on that face. In addition, we introduce CM as a centrality measure to help FM to reduce some of the errors from unnecessary areas such as the hair, neck or background. The VOC then accompanies the feature vectors in the EM process to estimate the optimal models by modified-GMM training. In the experiments, we introduce a real-world database, called KoFace, besides applying some public databases such as the Yale and the ORL database. The KoFace database is composed of 106 face subjects under diverse illumination effects including shadows and highlights. The results show that our proposed approach gives a higher Face Recognition Rate (FRR) than the GMM baseline for indoor and outdoor datasets in the real-world KoFace database (94% and 85%, respectively) and in ORL, Yale databases (97% and 100% respectively).

• Korean Journal of Construction Engineering and Management
• /
• v.10 no.3
• /
• pp.139-147
• /
• 2009

Accurate construction cost estimation in the initial stage of building project plays a key role for project success and for mitigation of disputes. Total construction cost(TCC) estimation of apartment projects in Vietnam has become more important because those projects increasingly rise in quantity with the urbanization and population growth. This paper presents the application of artificial neural networks(ANNs) in estimating TCC of apartment projects. Ninety-one questionnaires were collected to identify input variables. Fourteen data sets of completed apartment projects were obtained and processed for training and generalizing the neural network(NN). MATLAB software was used to train the NN. A program was constructed using Visual C++ in order to apply the neural network to realistic projects. The results suggest that this model is reasonable in predicting TCCs for apartment projects and reinforce the reliability of using neural networks to cost models. Although the proposed model is not validated in a rigorous way, the ANN-based model may be useful for both practitioners and researchers. It facilitates systematic predictions in early phases of construction projects. Practitioners are more proactive in estimating construction costs and making consistent decisions in initial phases of apartment projects. Researchers should benefit from exploring insights into its implementation in the real world. The findings are useful not only to researchers and practitioners in the Vietnam Construction Industry(VCI) but also to participants in other developing countries in South East Asia. Since Korea has emerged as the first largest foreign investor in Vietnam, the results of this study may be also useful to participants in Korea.

• Geomechanics and Engineering
• /
• v.20 no.2
• /
• pp.87-101
• /
• 2020

The construction of combined pile-raft foundations is considered as the main option in designing foundations in high-rise buildings, especially in soils close to the ground surface which do not have sufficient bearing capacity to withstand building loads. This paper deals with the geotechnical report of the Northern Fereshteh area of Tabriz, Iran, and compares the characteristics of the single pile foundation with the two foundations of pile group and geogrid. Besides, we investigate the effects of five principal parameters including pile diameter and length, the number of geogrid layers, the depth of groundwater level, and pore water pressure on vertical consolidation settlement and pore water pressure changes over a year. This study assessed the mechanism of the failure of the soil under the foundation using numerical analysis as well. Numerical analysis was performed using the two-dimensional finite element PLAXIS software. The results of fifty-four models indicate that the diameter of the pile tip, either as a pile group or as a single pile, did not have a significant effect on the reduction of the consolidation settlement in the soil in the Northern Fereshteh Street region. The optimum length for the pile in the Northern Fereshteh area is 12 meters, which is economically feasible. In addition, the construction of four-layered ten-meter-long geogrids at intervals of 1 meter beneath the deep foundation had a significant preventive impact on the consolidation settlement in clayey soils.

• Korean Journal of Computational Design and Engineering
• /
• v.13 no.2
• /
• pp.139-152
• /
• 2008

In order to survive worldwide competition, today's industries are experiencing strong pressure to introduce higher quality products with lower cost and shorter lead-time. Therefore, the role of design in the process of product development is increasing in significance. In this research, two methods for improving the design capability are proposed: construction of design environment and design automation using 3D CAD system. The designers and design process are the core of product design using 3D CAD system. In order to maximize the design performance, construction of the design environment including selection of a suitable system, designer training for best use of the system, establishment of an efficient design process, and stabilization of the environment are required. A method is suggested to construct design environment by systematizing the contents of the projects and consulting experiences carried out for various categories of business such as electronic devices, motorcycles, electricity parts, sanitary wares, injection molds, and die casing molds. Design automation helps reduce tedious and time-consuming jobs, simplify complicated and error-prone modeling and drawing works to shorten the lead time and improve the product quality. To develop a design automation system, understanding the process and the related knowledge on design are very important before implementing the system using API provided by 3D CAD system. In this research, an eight-step procedure is proposed for the development of a design automation system. These eight steps are analysis of needs, determination of specification, verification of specification using 3D CAD system, inspection of related API functions, programming, field test, application in practice, and maintenance. A case study in which five design automation systems in the design of turbine generators using the proposed method is introduced in detail. These systems play important roles in the generation of various output items including 3D models, drafts, material information, and NC data. The case study shows how effectively the design time is reduced and the quality improved using those systems.

• Journal of Korean Society on Water Environment
• /
• v.24 no.1
• /
• pp.91-98
• /
• 2008

This work was performed to develop a model possible to predict the influent flow and influent components, which are one of main disturbances causing process problems at the operation of municipal wastewater treatment plant. In this study, artificial neural network (ANN) was used in order to develop a model that was able to predict the influent flow, $COD_{Mn}$, SS, TN 1 day-ahead, 2day-ahead and 3 day ahead. Multi-layer feed-forward back-propagation network was chosen as neural network type, and tanh-sigmoid function was used as activation function to transport signal at the neural network. And Levenberg-Marquart (LM) algorithm was used as learning algorithm to train neural network. Among 420 data sets except missing data, which were collected between 2005 and 2006 at field plant, 210 data sets were used for training, and other 210 data sets were used for validation. As result of it, ANN model for predicting the influent flow and components 1-3day ahead could be developed successfully. It is expected that this developed model can be practically used as follows: Detecting the fault related to effluent concentration that can be happened in the future by combining with other models to predict process performance in advance, and minimization of the process fault through the establishment of various control strategies based on the detection result.

• Proceedings of the IEEK Conference
• /
• summer
• /
• pp.382-386
• /
• 2004

Recently, with the machine learning trend, most of the machine translation systems on over the world use two syntax tree sets of two relevant languages to learn syntactic tree transfer rules. However, for the English-Vietnamese language pair, this approach is impossible because until now we have not had a Vietnamese syntactic tree set which is correspondent to English one. Building of a very large correspondent Vietnamese syntactic tree set (thousands of trees) requires so much work and take the investment of specialists in linguistics. To take advantage from our available English-Vietnamese Corpus (EVC) which was tagged in word alignment, we choose the SITG (Stochastic Inversion Transduction Grammar) model to construct English- Vietnamese syntactic tree sets automatically. This model is used to parse two languages at the same time and then carry out the syntactic tree transfer. This English-Vietnamese bilingual syntactic tree set is the basic training data to carry out transferring automatically from English syntactic trees to Vietnamese ones by machine learning models. We tested the syntax analysis by comparing over 10,000 sentences in the amount of 500,000 sentences of our English-Vietnamese bilingual corpus and first stage got encouraging result $(analyzed\;about\;80\%)[5].$ We have made use the TBL algorithm (Transformation Based Learning) to carry out automatic transformations from English syntactic trees to Vietnamese ones based on that parallel syntactic tree transfer set[6].