• Information Systems Review
• /
• v.22 no.4
• /
• pp.185-203
• /
• 2020

With the advent of 4th industrial revolution, the manufacturing industry is converging with ICT and changing into the era of smart manufacturing. In the smart factory, all machines and facilities are connected based on ICT, and thus security should be further strengthened as it is exposed to complex security threats that were not previously recognized. To reduce the risk of security incidents and successfully implement smart factories, it is necessary to identify key security factors to be applied, taking into account the characteristics of the industrial environment of smart factories utilizing ICT. In this study, we propose a 'hierarchical classification model of security factors in smart factory' that includes terminal, network, platform/service categories and analyze the importance of security factors to be applied when developing smart factories. We conducted an assessment of importance of security factors to the groups of smart factories and security experts. In this study, the relative importance of security factors of smart factory was derived by using AHP technique, and the priority among the security factors is presented. Based on the results of this research, it contributes to building the smart factory more securely and establishing information security required in the era of smart manufacturing.

• Journal of Intelligence and Information Systems
• /
• v.24 no.1
• /
• pp.205-225
• /
• 2018

Convolutional Neural Network (ConvNet) is one class of the powerful Deep Neural Network that can analyze and learn hierarchies of visual features. Originally, first neural network (Neocognitron) was introduced in the 80s. At that time, the neural network was not broadly used in both industry and academic field by cause of large-scale dataset shortage and low computational power. However, after a few decades later in 2012, Krizhevsky made a breakthrough on ILSVRC-12 visual recognition competition using Convolutional Neural Network. That breakthrough revived people interest in the neural network. The success of Convolutional Neural Network is achieved with two main factors. First of them is the emergence of advanced hardware (GPUs) for sufficient parallel computation. Second is the availability of large-scale datasets such as ImageNet (ILSVRC) dataset for training. Unfortunately, many new domains are bottlenecked by these factors. For most domains, it is difficult and requires lots of effort to gather large-scale dataset to train a ConvNet. Moreover, even if we have a large-scale dataset, training ConvNet from scratch is required expensive resource and time-consuming. These two obstacles can be solved by using transfer learning. Transfer learning is a method for transferring the knowledge from a source domain to new domain. There are two major Transfer learning cases. First one is ConvNet as fixed feature extractor, and the second one is Fine-tune the ConvNet on a new dataset. In the first case, using pre-trained ConvNet (such as on ImageNet) to compute feed-forward activations of the image into the ConvNet and extract activation features from specific layers. In the second case, replacing and retraining the ConvNet classifier on the new dataset, then fine-tune the weights of the pre-trained network with the backpropagation. In this paper, we focus on using multiple ConvNet layers as a fixed feature extractor only. However, applying features with high dimensional complexity that is directly extracted from multiple ConvNet layers is still a challenging problem. We observe that features extracted from multiple ConvNet layers address the different characteristics of the image which means better representation could be obtained by finding the optimal combination of multiple ConvNet layers. Based on that observation, we propose to employ multiple ConvNet layer representations for transfer learning instead of a single ConvNet layer representation. Overall, our primary pipeline has three steps. Firstly, images from target task are given as input to ConvNet, then that image will be feed-forwarded into pre-trained AlexNet, and the activation features from three fully connected convolutional layers are extracted. Secondly, activation features of three ConvNet layers are concatenated to obtain multiple ConvNet layers representation because it will gain more information about an image. When three fully connected layer features concatenated, the occurring image representation would have 9192 (4096+4096+1000) dimension features. However, features extracted from multiple ConvNet layers are redundant and noisy since they are extracted from the same ConvNet. Thus, a third step, we will use Principal Component Analysis (PCA) to select salient features before the training phase. When salient features are obtained, the classifier can classify image more accurately, and the performance of transfer learning can be improved. To evaluate proposed method, experiments are conducted in three standard datasets (Caltech-256, VOC07, and SUN397) to compare multiple ConvNet layer representations against single ConvNet layer representation by using PCA for feature selection and dimension reduction. Our experiments demonstrated the importance of feature selection for multiple ConvNet layer representation. Moreover, our proposed approach achieved 75.6% accuracy compared to 73.9% accuracy achieved by FC7 layer on the Caltech-256 dataset, 73.1% accuracy compared to 69.2% accuracy achieved by FC8 layer on the VOC07 dataset, 52.2% accuracy compared to 48.7% accuracy achieved by FC7 layer on the SUN397 dataset. We also showed that our proposed approach achieved superior performance, 2.8%, 2.1% and 3.1% accuracy improvement on Caltech-256, VOC07, and SUN397 dataset respectively compare to existing work.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.48 no.2
• /
• pp.66-77
• /
• 2011

We propose a framework for grouping photographic advertising images that employs a hierarchical indexing scheme based on efficient feature combinations. The study provides one specific application of effective tools for monitoring photographic advertising information through online and offline channels. Specifically, it develops a preprocessor for advertising image information tracking. We consider both global features that contain general information on the overall image and local features that are based on local image characteristics. The developed local features are invariant under image rotation and scale, the addition of noise, and change in illumination. Thus, they successfully achieve reliable matching between different views of a scene across affine transformations and exhibit high accuracy in the search for matched pairs of identical images. The method works with global features in advance to organize coarse clusters that consist of several image groups among the image data and then executes fine matching with local features within each cluster to construct elaborate clusters that are separated by identical image groups. In order to decrease the computational time, we apply a conventional clustering method to group images together that are similar in their global characteristics in order to overcome the drawback of excessive time for fine matching time by using local features between identical images.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.31 no.6_1
• /
• pp.431-437
• /
• 2013

Landslides are caused by earthquakes or heavy rains. Recently the incidence of landslides has been increased. However, it is impossible to predict the occurrence of landslide exactly. The purpose of this research is that subdivide the classes of elements in the landslide management system by using spatial analysis technique and AHP method. The existing landslide management system is only comprised of weighted value the slope elements without weighted value about the slop direction elements. For the accuracy improvement in landslide occurrence point, weighted value about the slope direction should be considered. This research is focused on segmentation in slope direction three categories. If the direction of landslide does not affect the structure, I do not think the subject is worth considerating. Based on these results will discuss the improvements in Landslides management systems. Analysis results, segmentation on the slope and the slope direction are needed. Segmented categories about topography elements will be increase the accuracy of landslides management system. Also, since topography of the elements is only considered, segmentation of different elements is needed.

• Science of Emotion and Sensibility
• /
• v.13 no.2
• /
• pp.381-390
• /
• 2010

The automotive seat comfort evaluation was to begin a key aspect in seat design. It depended largely on the basic mechanical aspect such as geometric parameters of seat, choice of suspension system and cushion material used. Until recently, seat comfort evaluation advanced to evaluate subjective sensitivity of human. The external literatures showed in the last decade, there have been very few attempts to establish and document automotive seat comfort evaluation. In 2006 Smith, D. proposed the statistically reliable tool in giving a numeric rating for set comfort and the tool was used in the many country. On the other hands, we, in Korea, had not the reliable tool for the automotive seat comfort evaluation. So that, we studied to develope the questionnaire for seat comfort evaluation based on Smith, D.(2006) and some studies. As a result, we developed 36 contents for the automotive seat comfort evaluation with the help of professional in Korean automotive industry. Here, 36 contents were identified as the dimensions that represent the human sensitivity and psychological feeling on comfortable seat. Also, we derived the priorities for the 36 contents by using analytic hierarchy process (AHP), based on the judgments of 30 experts and drivers. This study will help the designers and developers clarify the conceptual and abstract aspect of the design evaluation by proposing a more systematic and process-oriented method.

• Journal of Technology Innovation
• /
• v.22 no.4
• /
• pp.145-164
• /
• 2014

Nuclear fusion is one of the most promising options for generating large amounts of carbon-free energy in the future. Major countries such as China, EU, and Japan have established a national plan for DEMO construction and they are implementing it. Korea has started a nuclear fusion research and development by the KSTAR project started in 1995. There are matured needs for a full-scale research and development initiatives to ensure competition with the major countries for DEMO as well as achieve the final goal to commercialize fusion energy. In this paper, we apply the TRL and AHP methods in order to identify the key technologies to conduct DEMO R&D. We propose the priorities of future R&D on DEMO by deriving a core technology in the field. At first, we review the scientific theory of fusion and trend of progress of DEMO activities in major countries. For previous studies, we review TRL and AHP methods to examine the technology classification system of DEMO and identify key technologies. We apply TRL method to identify readiness level of DEMO technologies and AHP to compensate shortcoming of TRL. The key technologies of DEMO to be secured from a synthesis result of the TRL and AHP are burning plasma, plasma facing material, structural material, high frequency heating, neutral particle beam, safety, plasma diagnostic, and simulation technologies.

• Korean journal of applied entomology
• /
• v.56 no.1
• /
• pp.29-39
• /
• 2017

To compare the features of ground-dwelling invertebrates according agroecosystems, we selected paddy fields, dry fields, orchards in the Eastern region of Korea. The surveys were performed by using pit-fall traps twice per year from 2013 to 2015. Total 6,420 individuals of 172 species belonging to 13 orders, 58 families were investigated in the Eastern region, the species of Hymenoptera (38.26%), Orthoptera (16.28%) accounted large portion of the communities. In the geographical observation, invertebrates were caught was 2,983 individuals in Gyeongsangnam-do, the diversity index of Gyeongsangbuk-do community was higher than of the others and abundance and species richness of paddy field were higher than from dry field or orchard. To understand the relation between taxonomic groups and environmental factors, we carried out the canonical correspondence analysis and hierarchical clustering. As a result, Homoptera, Blattaria, Isoptera, and Coleoptera were positively related to soil pH, soil temperature, and moisture contents, and negatively related to the others. Invertebrate community also were patterned dependently by type of ecosystems. This results were shown that distribution of invertebrates is a few influenced the relationship of the space habituated invertebrates and environmental factors.

• Journal of Plant Biotechnology
• /
• v.45 no.3
• /
• pp.242-249
• /
• 2018

This study was conducted to develop an SNP set that can be useful for marker-assisted breeding (MAB) in watermelon (Citrullus. lanatus L) using Genotyping-by-sequencing (GBS) analysis of 20 commercial elite watermelon inbreds. The result of GBS showed that 77% of approximately 1.1 billion raw reads were mapped on the watermelon genome with an average mapping region of about 4,000 Kb, which indicated genome coverage of 2.3%. After the filtering process, a total of 2,670 SNPs with an average depth of 31.57 and the PIC (Polymorphic Information Content) value of 0.1~0.38 for 20 elite inbreds were obtained. Among those SNPs, 55 SNPs (5 SNPs per chromosome that are equally distributed on each chromosome) were selected. For the understanding genetic relationship of 20 elite inbreds, PCA (Principal Component Analysis) was carried out with 55 SNPs, which resulted in the classification of inbreds into 4 groups based on PC1 (52%) and PC2 (11%), thus causing differentiation between the inbreds. A similar classification pattern for PCA was observed from hierarchical clustering analysis. The SNP set developed in this study has the potential for application to cultivar identification, F1 seed purity test, and marker-assisted backcross (MABC) not only for 20 elite inbreds but also for diverse resources for watermelon breeding.

• Korean Journal of Plant Taxonomy
• /
• v.43 no.4
• /
• pp.245-251
• /
• 2013

Paeonia lactiflora is a valuable natural resource for horticulture and traditional Chinese medicine. To propose conservation strategy and future utility of the wild Paeonia lactiflora populations recently found around the Gyeongju National Park, genetic diversity analysis using microsatellite markers were performed. Three populations in and near the Gyeongju N.P. and one population from Jilin, China were analyzed for five microsatellite markers, producing 61 alleles with mean observed heterozygosity($H_o$) of 0.452. $F_{ST}$ value (0.11642) suggested moderate level of genetic differentiation among the populations, and hierarchical AMOVA suggested most of the genetic variation resides within/among the individuals rather than among-population. While AMOVA with $F_{ST}$ suggested lack of genetic differentiation between the regional (Korean vs. Chinese) populations, AMOVA with $R_{ST}$, which incorporates the allele sizes, suggested considerable differentiation between them, but without significant statistical support. STRUCTURE analysis also suggested segregation of regional populations with presence of gene flow among the three Gyeongju N.P. populations. Considering small population size and scarcity of mature individuals, further protection and long-term monitoring are needed.

• Archives of design research
• /
• v.15 no.1
• /
• pp.191-202
• /
• 2002

As family types have been diverse, patterns of living and living space became diverse as much as users are. Therefore, it is needed to provide various remodeled design of living space corresponding to changes of users'living patterns, and to provide these remodeling process to users directly on the web. In this paper, use scenario for the Internet-based Virtual Apartment Remodeling Simulator is researched as an export system to remodel space in accordance with users diverse lifestyle paradigm and the website is developed. The study consists of four parts. First, the general concept of remodeling, including the range and types of remodeling, are defined, and the misleading terms in this field are reviewed and organized by secondary research Second, fixed factors and variable factors are differentiated in the complex building for residence and business that was decided as a basic building type in this study. Third, there needed a database for consulting, final material, pre-estimation real estimation for simulation of remodeling. This database was introduced along with floor plan and elevation. Finally, the remodeling simulator is presented by the case study developed on the web. The system structure and use scenario are also presented. In order to present and inspect design alternatives, prototype was produced. The Final simulator was enhanced by defeating problems regarding interface efficiency and missing information of existing online site.