• Journal of Software Engineering Society
• /
• v.28 no.1
• /
• pp.23-28
• /
• 2019

Recently, in order to effectively test deep neural network model for image processing application, researches have actively conducted to automatically generate data in corner-case that is not correctly predicted by the model. This paper proposes test data generation method that selects arbitrary words from input of system and transforms them into synonyms in order to test the bug reporter automatic assignment system based on sentence classification deep neural network model. In addition, we compare and evaluate the case of using proposed test data generation and the case of using existing difference-inducing test data generations based on various neuron coverages.

• The Journal of Information Systems
• /
• v.31 no.1
• /
• pp.115-146
• /
• 2022

Purpose The purpose of this study is to find meaningful insights using regression analysis to resolve the digital divide between generations. In the analysis process of this study, social network analysis was applied to approach it with a perspective differentiated from the existing statistical techniques. Design/methodology/approach This study used a social network analysis methodology that transforms and analyzes government-led survey data into relational data. First, the cross-sectional data were converted into relational data, and a continuous core-periphery model and multidimensional scaling method were applied. Afterwards, the relationship between various factors affecting the digital divide and the difference in influence were analyzed by generation. Findings According to the network analysis results, it can be seen that all generations commonly use 'information and news search' and 'living information service'. However, it can be seen that the centrally used services of each generation are clearly different from each other, and the degree of linkage between the services is also clearly different. In addition, it can be seen that the relationship between factors influencing the digital divide by generation is also different.

• Transactions on Electrical and Electronic Materials
• /
• v.18 no.2
• /
• pp.111-118
• /
• 2017

The increased diversity of different types of energy sources requires moving towards smart distribution networks. This paper proposes a probabilistic DG (distributed generation) units planning model to determine technology type, capacity and location of DG units while simultaneously allocating ESS (energy storage systems) based on pre-determined capacities. This problem is studied in a wind integrated power system considering loads, prices and wind power generation uncertainties. A suitable method for DG unit planning will reduce costs and improve reliability concerns. Objective function is a cost function that minimizes DG investment and operational cost, purchased energy costs from upstream networks, the defined cost to reliability index, energy losses and the investment and degradation costs of ESS. Electrical load is a time variable and the model simulates a typical radial network successfully. The proposed model was solved using the DICOPT solver under GAMS optimization software.

• The Journal of the Korea Contents Association
• /
• v.4 no.4
• /
• pp.107-112
• /
• 2004

As the internet service evolves fast recently, guarantee request of QoS (Quality of Service) by characteristic of traffic source as well as high rate of data is going greatest. Accordingly intemet service technology also is changing rapidly, technology that guarantee QoS in existent network service technology is developed or network model that guarantee new QoS is presented. By DiffServ-aware-MPLS network that present in IETF (Internet Engineering Task Force) to guarantee QoS in this treatise does comparative analysis with existent network model, relative show that is superior, and present direction that compose next generation network wish to.

• Transactions of the Korean hydrogen and new energy society
• /
• v.31 no.5
• /
• pp.436-443
• /
• 2020

The performance prediction model of a solid oxide fuel cell stack has been developed using deep neural network technique, one of the machine learning methods. The machine learning has been received much interest in various fields, including energy system mo- deling. Using machine learning technique can save time and cost requried in developing an energy system model being compared to the conventional method, that is a combination of a mathematical modeling and an experimental validation. Results reveal that the mean average percent error, root mean square error, and coefficient of determination (R2) range 1.7515, 0.1342, 0.8597, repectively, in maximum. To improve the predictability of the model, the pre-processing is effective and interpolative machine learning and application is more accurate than the extrapolative cases.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.7
• /
• pp.914-924
• /
• 2020

In this study we developed artificial network generation model, which can generate on-line social network. The suggested model can represent not only scale-free and small-world properties, but also can produce networks with various values of topological characteristics through controlling two input parameters. For this purpose, two parameter K and P are introduced, K for controlling the strength of preferential attachment and P for controlling clustering coefficient. It is found out on-line social network can be generated with the combinations of K(0~10) and P(0.3~0.5) or K=0 and P=0.9. Under these combinations of P and K small-world and scale-free properties are well represented. Node degree distribution follows power-law. Clustering coefficients are between 0.130 and 0.238, and average shortest path distance between 5.641 and 5.985. It is also found that on-line social network properties are maintained as network node size increases from 5,000 to 10,000.

• Journal of Korea Society of Industrial Information Systems
• /
• v.18 no.6
• /
• pp.39-49
• /
• 2013

In recent, many countries and research institutes have been studying how to construct the high-speed subscriber access network. Task Force team of IEEE 802.3ah has accomplished the standardization of EPON which is the next generation subscriber access network. EPON doesn't still have the bandwidth enough to support the new service(e.g various multimedia service) which demands the high bandwidth. For these new multimedia services,10G EPON is the next generation subscriber access network which expanded the up-down bandwidth range of 1G EPON 10 times in order to support demanding high bandwidth. We have proposed the model which can accommodate IEEE 802.1 AVB traffics smoothly in 10G EPON and suggesting the Intra-ONU scheduling model which makes this model operate effectively.

• The Transactions of the Korean Institute of Electrical Engineers D
• /
• v.54 no.9
• /
• pp.567-569
• /
• 2005

NARX(Nonlinear AutoRegressive with eXogenous input) neural network was used for prediction of nuclear reactor behavior which was influenced by control rods in short-term period and also by the concentration of xenon and boron in long-term period in load following operations. The developed model was designed to predict reactor power, xenon worth and axial offset with different burnup states when control rods and boron were adjusted in load following operations. Data of the Korea Next Generation PWR were collected by ONED94 code. The test results presented exhibit the capability of the NARX neural network model to capture the long term and short term dynamics of the reactor core and the developed model seems to be utilized as a handy tool for the use of a plant simulation.

• Communications for Statistical Applications and Methods
• /
• v.30 no.6
• /
• pp.589-603
• /
• 2023

Deep generative models target to infer the underlying true data distribution, and it leads to a huge success in generating fake-but-realistic data. Regarding such a perspective, the data attributes can be a crucial factor in the data generation process since non-existent counterfactual samples can be generated by altering certain factors. For example, we can generate new portrait images by flipping the gender attribute or altering the hair color attributes. This paper proposes counterfactual disentangled variational autoencoder generative adversarial networks (CDVAE-GAN), specialized for data attribute level counterfactual data generation. The structure of the proposed CDVAE-GAN consists of variational autoencoders and generative adversarial networks. Specifically, we adopt a Gaussian variational autoencoder to extract low-dimensional disentangled data features and auxiliary Bernoulli latent variables to model the data attributes separately. Also, we utilize a generative adversarial network to generate data with high fidelity. By enjoying the benefits of the variational autoencoder with the additional Bernoulli latent variables and the generative adversarial network, the proposed CDVAE-GAN can control the data attributes, and it enables producing counterfactual data. Our experimental result on the CelebA dataset qualitatively shows that the generated samples from CDVAE-GAN are realistic. Also, the quantitative results support that the proposed model can produce data that can deceive other machine learning classifiers with the altered data attributes.

• KIPS Transactions on Software and Data Engineering
• /
• v.6 no.4
• /
• pp.203-210
• /
• 2017

In this paper, we propose an effective neural network model for image caption generation and model transfer. This model is a kind of multi-modal recurrent neural network models. It consists of five distinct layers: a convolution neural network layer for extracting visual information from images, an embedding layer for converting each word into a low dimensional feature, a recurrent neural network layer for learning caption sentence structure, and a multi-modal layer for combining visual and language information. In this model, the recurrent neural network layer is constructed by LSTM units, which are well known to be effective for learning and transferring sequence patterns. Moreover, this model has a unique structure in which the output of the convolution neural network layer is linked not only to the input of the initial state of the recurrent neural network layer but also to the input of the multimodal layer, in order to make use of visual information extracted from the image at each recurrent step for generating the corresponding textual caption. Through various comparative experiments using open data sets such as Flickr8k, Flickr30k, and MSCOCO, we demonstrated the proposed multimodal recurrent neural network model has high performance in terms of caption accuracy and model transfer effect.