• Journal of Intelligence and Information Systems
• /
• v.24 no.3
• /
• pp.1-19
• /
• 2018

Large amount of data is now available for research and business sectors to extract knowledge from it. This data can be in the form of unstructured data such as audio, text, and image data and can be analyzed by deep learning methodology. Deep learning is now widely used for various estimation, classification, and prediction problems. Especially, fashion business adopts deep learning techniques for apparel recognition, apparel search and retrieval engine, and automatic product recommendation. The core model of these applications is the image classification using Convolutional Neural Networks (CNN). CNN is made up of neurons which learn parameters such as weights while inputs come through and reach outputs. CNN has layer structure which is best suited for image classification as it is comprised of convolutional layer for generating feature maps, pooling layer for reducing the dimensionality of feature maps, and fully-connected layer for classifying the extracted features. However, most of the classification models have been trained using online product image, which is taken under controlled situation such as apparel image itself or professional model wearing apparel. This image may not be an effective way to train the classification model considering the situation when one might want to classify street fashion image or walking image, which is taken in uncontrolled situation and involves people's movement and unexpected pose. Therefore, we propose to train the model with runway apparel image dataset which captures mobility. This will allow the classification model to be trained with far more variable data and enhance the adaptation with diverse query image. To achieve both convergence and generalization of the model, we apply Transfer Learning on our training network. As Transfer Learning in CNN is composed of pre-training and fine-tuning stages, we divide the training step into two. First, we pre-train our architecture with large-scale dataset, ImageNet dataset, which consists of 1.2 million images with 1000 categories including animals, plants, activities, materials, instrumentations, scenes, and foods. We use GoogLeNet for our main architecture as it has achieved great accuracy with efficiency in ImageNet Large Scale Visual Recognition Challenge (ILSVRC). Second, we fine-tune the network with our own runway image dataset. For the runway image dataset, we could not find any previously and publicly made dataset, so we collect the dataset from Google Image Search attaining 2426 images of 32 major fashion brands including Anna Molinari, Balenciaga, Balmain, Brioni, Burberry, Celine, Chanel, Chloe, Christian Dior, Cividini, Dolce and Gabbana, Emilio Pucci, Ermenegildo, Fendi, Giuliana Teso, Gucci, Issey Miyake, Kenzo, Leonard, Louis Vuitton, Marc Jacobs, Marni, Max Mara, Missoni, Moschino, Ralph Lauren, Roberto Cavalli, Sonia Rykiel, Stella McCartney, Valentino, Versace, and Yve Saint Laurent. We perform 10-folded experiments to consider the random generation of training data, and our proposed model has achieved accuracy of 67.2% on final test. Our research suggests several advantages over previous related studies as to our best knowledge, there haven't been any previous studies which trained the network for apparel image classification based on runway image dataset. We suggest the idea of training model with image capturing all the possible postures, which is denoted as mobility, by using our own runway apparel image dataset. Moreover, by applying Transfer Learning and using checkpoint and parameters provided by Tensorflow Slim, we could save time spent on training the classification model as taking 6 minutes per experiment to train the classifier. This model can be used in many business applications where the query image can be runway image, product image, or street fashion image. To be specific, runway query image can be used for mobile application service during fashion week to facilitate brand search, street style query image can be classified during fashion editorial task to classify and label the brand or style, and website query image can be processed by e-commerce multi-complex service providing item information or recommending similar item.

• Journal of Intelligence and Information Systems
• /
• v.27 no.3
• /
• pp.57-73
• /
• 2021

Maintenance and prevention of failure through anomaly detection of ICT infrastructure is becoming important. System monitoring data is multidimensional time series data. When we deal with multidimensional time series data, we have difficulty in considering both characteristics of multidimensional data and characteristics of time series data. When dealing with multidimensional data, correlation between variables should be considered. Existing methods such as probability and linear base, distance base, etc. are degraded due to limitations called the curse of dimensions. In addition, time series data is preprocessed by applying sliding window technique and time series decomposition for self-correlation analysis. These techniques are the cause of increasing the dimension of data, so it is necessary to supplement them. The anomaly detection field is an old research field, and statistical methods and regression analysis were used in the early days. Currently, there are active studies to apply machine learning and artificial neural network technology to this field. Statistically based methods are difficult to apply when data is non-homogeneous, and do not detect local outliers well. The regression analysis method compares the predictive value and the actual value after learning the regression formula based on the parametric statistics and it detects abnormality. Anomaly detection using regression analysis has the disadvantage that the performance is lowered when the model is not solid and the noise or outliers of the data are included. There is a restriction that learning data with noise or outliers should be used. The autoencoder using artificial neural networks is learned to output as similar as possible to input data. It has many advantages compared to existing probability and linear model, cluster analysis, and map learning. It can be applied to data that does not satisfy probability distribution or linear assumption. In addition, it is possible to learn non-mapping without label data for teaching. However, there is a limitation of local outlier identification of multidimensional data in anomaly detection, and there is a problem that the dimension of data is greatly increased due to the characteristics of time series data. In this study, we propose a CMAE (Conditional Multimodal Autoencoder) that enhances the performance of anomaly detection by considering local outliers and time series characteristics. First, we applied Multimodal Autoencoder (MAE) to improve the limitations of local outlier identification of multidimensional data. Multimodals are commonly used to learn different types of inputs, such as voice and image. The different modal shares the bottleneck effect of Autoencoder and it learns correlation. In addition, CAE (Conditional Autoencoder) was used to learn the characteristics of time series data effectively without increasing the dimension of data. In general, conditional input mainly uses category variables, but in this study, time was used as a condition to learn periodicity. The CMAE model proposed in this paper was verified by comparing with the Unimodal Autoencoder (UAE) and Multi-modal Autoencoder (MAE). The restoration performance of Autoencoder for 41 variables was confirmed in the proposed model and the comparison model. The restoration performance is different by variables, and the restoration is normally well operated because the loss value is small for Memory, Disk, and Network modals in all three Autoencoder models. The process modal did not show a significant difference in all three models, and the CPU modal showed excellent performance in CMAE. ROC curve was prepared for the evaluation of anomaly detection performance in the proposed model and the comparison model, and AUC, accuracy, precision, recall, and F1-score were compared. In all indicators, the performance was shown in the order of CMAE, MAE, and AE. Especially, the reproduction rate was 0.9828 for CMAE, which can be confirmed to detect almost most of the abnormalities. The accuracy of the model was also improved and 87.12%, and the F1-score was 0.8883, which is considered to be suitable for anomaly detection. In practical aspect, the proposed model has an additional advantage in addition to performance improvement. The use of techniques such as time series decomposition and sliding windows has the disadvantage of managing unnecessary procedures; and their dimensional increase can cause a decrease in the computational speed in inference.The proposed model has characteristics that are easy to apply to practical tasks such as inference speed and model management.

• Journal of the Korean Institute of Landscape Architecture
• /
• v.44 no.2
• /
• pp.83-95
• /
• 2016

Strategies for parks and open spaces in the 21st century have moved from focusing on specific elements, such as quantitative growth and ecological and recreational aspects, to green infrastructure, which refers to a multi-functional network of open and green spaces offering a range of benefits. In the case of London, green infrastructure is realised as an integral part of urban infrastructure, involving physical and social infrastructure as well as practical spatial planning at the local level within statutory urban planning as part of a continuously developing green infrastructure framework with a theoretical basis. Taking this perspective, the present study looks at alterations to and developments in green infrastructure policies in the London Plan, the green grid framework as detailed in the city's strategic implementation of green infrastructure. Various trends and characteristics of the policies adopted in the London Plan and some implications are deduced, with three main results being identified. The first is a clear division of roles among the national government, Greater London Authority and borough councils, with local plans established under the guidance of the National Planning Policy Framework (NPPF) and the London Plan. Green infrastructure policies in the London Plan have been applied at a high rate in the boroughs' local plans, which leads to another, linked point. Secondly, green infrastructure policies and the green grid as an implementation framework have been consistently extended and developed through consolidating the London Plan, despite the change of government. Finally, in order to achieve the London Plan, the Mayor of London implemented policies by partnership and supporting programmes for London boroughs. Recently, the Seoul Metropolitan Authority introduced a parks and green spaces development policy, but the London case remains a good example; this is because green infrastructure policies in London were not a manifesto pledge but rather have been continuously and consistently advanced regardless of party politics and thus realised as long-term planning.

• Archives of design research
• /
• v.20 no.3 s.71
• /
• pp.151-162
• /
• 2007

How can new media devices and networks provide an effective response to the world's growing sector of cultural and historically-minded travelers? This study emerged from the question of how mobile handsets can change the nature of cultural and historical tourism in ubiquitous city environments. As wireless network and mobile IT have rapidly developed, it becomes possible to deliver cultural and historical information on the site through mobile handset as a tour guidance system. The paper describes the development of a new type of mobile tourism platform for site-specific cultural and historical information. The central objective of the project was to organize this cultural and historical walking tour around the mobile handset and its unique advantages (i.e. portability, multi-media capacity, access to wireless internet, and location-awareness potential) and then integrate the tour with a historical story and role-playing game that would deepen the mobile user's interest in the sites being visited, and enhance his or her overall experience of the area. The project was based on twelve locations that were culturally and historically significant to Korean War era in Busan. After the mobile tour game prototype was developed for this route, it was evaluated at the 10th PIFF (Pusan International Film Festival). After use test, some new strategies for developing mobile "edutainment content" to deliver cultural historical contents of the location were discussed. Combining 'edutainment' with a cultural and historical mobile walking tour brings a new dimension to existing approaches of the tourism and mobile content industry.

• Geotechnical Engineering
• /
• v.13 no.3
• /
• pp.101-122
• /
• 1997

A spiting reinforcement system is composed of a series of radially installed reinforcing spites along the perimeter of the tunnel opening ahead of excavation. The reinforcing spill network is extended into the in-situ soil mass both radially and longitudinally The sailing reinforcement system has been successfully used for the construction of underground openings to reinforce weak rock formations on several occasions. The application of this spiting reinforcement system is currently extended to soft ground tunneling in limited occasions because of lack of reliable analysis and design methods. A method of threetimensional limit equilibrium stability analysis of the smile-reinforced shallow tunnel in soft ground is presented. The shape of the potential failure wedge for the case of smile-reinforced shallow tunnel is assumed on the basis of the results of three dimensional finite element analyses. A criterion to differentiate the spill-reinforced shallow tunnel from the smile-reinforced deep tunnel is also formulated, where the tunnel depth, soil type, geometry of the tunnel and reinforcing spites, together with soil arching effects, are considered. To examine the suitability of the proposed method of threedimensional stability analysis in practice, overall stability of the spill-reinforced shallow tunnel at facing is evaluated, and the predicted safety factors are compared with results from twotimensional analyses. Using the proposed method of threetimensional limit equilibrium stability analysis of the smile-reinforced shallow tunnel in soft ground, a parametric study is also made to investigate the effects of various design parameters such as tunnel depth, smile length and wadial spill spacing. With slight modifications the analytical method of threeiimensional stability analysis proposed may also be extended for the analysis and design of steel pipe reinforced multi -step grouting technique frequently used as a supplementary reinforcing method in soft ground tunnel construction.

• Journal of Soil and Groundwater Environment
• /
• v.12 no.5
• /
• pp.39-53
• /
• 2007

Generally, fractured medium can be described with some key parameters, such as hydraulic conductivities or random field of hydraulic conductivities (continuum model), spatial and statistical distribution of permeable fractures (discrete fracture network model). Investigating the practical applicability of the well-known conceptual models for the description of groundwater flow in fractured media, various types of hydraulic tests were applied to studies on the highly fractured media in Geumsan, Korea. Results from single-hole packer test show that the horizontal hydraulic conductivities in the permeable media are between $7.67{\times}10^{-10}{\sim}3.16{\times}10^{-6}$ m/sec, with $7.70{\times}10^{-7}$ m/sec arithmetic mean and $2.16{\times}10^{-7}$ m/sec geometric mean. Total number of test interval is 110 at 8 holes. The number of completely impermeable interval is 9, and the low permeable interval - below $1.0{\times}10^{-8}$ m/sec is 14. In other words, most of test intervals are permeable. The vertical distribution of hydraulic conductivities shows apparently the good correlation with the results of flowmeter test. But the results from the cross-hole test show some different features. The results from the cross-hole test are highly related to the connectivity and/or the binary properties of fractured media; permeable and impermeable. From the viewpoint of the connection, the application of the general stochastic approach with a single continuum model may not be appropriate even in the moderately or highly permeable fractured medium. Then, further studies on the investigation method and the analysis procedures should be required for the reasonable and practical design of the conceptual model, with which the binary properties, including permeable/impermeable features, can be described.

• Journal of the Korean Institute of Landscape Architecture
• /
• v.36 no.6
• /
• pp.43-54
• /
• 2009

This study attempted to develop a model for selecting sites for ecologically effective, multi-functional wetlands during the environmental and ecological planning stage, prior to land use Planning. This model was developed with an emphasis upon the creation of a water circulation system for a newly-created city, dispersing and retaining the run-off that is increased due to urbanization and securing spaces to create wetlands that can promote urban biodiversity. A series of Precesses for selecting sites for wetland restoration and creation - watershed analysis, selection of evaluation items, calculation of weights, reparation of thematic maps and synthesis - were incorporated into the model. Its potentials and limitations were examined by applying it to the recently-planned WiRae New Community Development Area, which is located in the Seoul metropolitan region. At the watershed analysis stage, the site was divided into 13 sub-catchment areas. Inflow to watersheds including the area was $3,020,765m^3$ Run-off before and after development is estimated as $1,901,969m^3$ and $1,970,735{\sim}2,039,502m^3$, respectively. The total storage capacity required in the development area amounts to $68,766{\sim}137,533m^3$. When thematic maps were overlapped during the selection stage for wetland sites, 13 sub-catchment areas were prioritized for wetland restoration and creation. The locations and areas for retaining run-off showed that various types of wetlands, including retaining wetlands (area wetlands), riverine wetlands (linear wetlands) and pond wetlands (point wetlands), can be created and that they can be systematically connected. By providing a basic framework for the water circulation system plan of an entire city, it may be used effectively in the space planning stage, such as planning an urban eco-network through integration with greet areas. In order to estimate reasonable run-off and create an adequate water circulation system however, a feedback process following land use planning is required. This study strived to promote urban changes in a positive direction while minimizing urban changes in negative forms.

• Journal of Distribution Research
• /
• v.15 no.5
• /
• pp.155-174
• /
• 2010

In many cases, street vendors are means of living and/or a solution to unemployment for low income group. Street vendors have both of positive and negative effects on traditional market revitalization. This two-sided perspective of the vendors has produced different views of interests groups. In this study, the features of street vendors are examined to present related issues, and a survey result conducted on interests groups is presented to offer effective management measures. As traditional management about street vendors are focused on crackdown, relationship between street traders and government has been not very much mutually cooperative, and at times some traders organized groups who are against government policies. With the premise admitting street vendors as one element of distribution network, it is possible to access the management measures for street vendors in organizational, quantitative or qualitative perspectives. However, I believe it is not recommendable to enforce multi-perspective approach at a time. It is because street traders still have quite strong animosity against government policies. Therefore, serious misunderstanding and side-effects on our society could be brought if the government makes hasty and forcible attempts to legalize street vendors. In political position, overreaching actions of government could hardly produce positive results because policy making and its enforcement need each of timeliness. In a similar way, government's policies for street vendors need to come into effect gradually. Management measures for street vendors can come in short-, mid- and long-term. In short-term, government should try to reduce animosity of street traders along with minimizing institutional and political pressure on them. As a mid-term solution, plans to bring vendors over to institutional boundaries by improving them are required. Last but not least, in the long term, government should design policies which are to help street vendors settle down and maintain successfully in the boundaries. Besides, policies related to street vendors need to come in effect in a way that closely connected to interests groups and businesses because those policies would get involve many interests groups and businesses in diverse perspectives.

• Korean Journal of Environment and Ecology
• /
• v.21 no.3
• /
• pp.243-256
• /
• 2007

The purpose of this paper is aimed at identifying the planting condition of greenbelt axis covering forests located at Sinhyeon-eup, Geoje-si and also establishing improvement plans for the ecological organization. The study was executed with buffer green space designed to mitigate noise, which is located at a halfway point linking Mt. Yukkyo(altitude of 50m) and Mt. Jungmae(altitude of 131m). The number of the biotope patterns was classified into 17 in total: the two are urbanized districts such as a townified district and streets and another 15 are greenbelts and open space such as forest biotope, inland water biotope, and landscaping tree plantations biotope. According to the analysis of biotope types, it was estimated that the making use of already established buffer greens as a linking medium with a foothold of Mt. Yukkyo and Jungmae, whose natural eco-system is well suited for habitation of living organism, is the one and only way to the influx of living organism into the downtown area. The green coverage rate of the base green area, sub-base green area and linkage green area was 160.29%, 128.37% and 44.37% respectively; the green capacity coefficient(i.e. GVZ[$Gr{\ddot{u}}nvolumezahl$]) for base green area, sub-base green area and linkage green area was $4.04m^3/m^2,\;3.95m^3/m^2\;and\;0.65m^3/m^2$ respectively. The base green area has constituted multi-layered vegetation structure and thus played a role as habitats for living organism and supply centre of species, whereas the sub-base green area has destroyed lower layer vegetation, and the linkage green area was in bad shape due to the lack of planting volume and damage of the shrub layer. Accordingly, this research paper intended to suggest detailed implementation plans for the improvement in landscape for city dwellers' use and relaxation; in other words, this paper focused on ecological build-up for the Influx of wild birds into the downtown area for the promotion of bio-diversity of species through the linkage of base green areas and the fostering of nature observing trails for citizens as well as the connecting of green areas through the build-up of roadside greens to make these green areas to be efficiently used as corridors for the influx of wild birds and bio-organism habitation and for citizens' using space.

• Particle and aerosol research
• /
• v.11 no.3
• /
• pp.63-75
• /
• 2015

Atmospheric environment in Korea is influenced by outside, especially China. The concentrations of air pollutants in China have showed decreasing trends since 2000. However, these concentration levels in China are still higher than other developed countries. The Chinese Government has tried several measures to control the air pollution. In this study, the details of the amendments and smog reduction strategies in China, especially for Beijing are reviewed and the strategies for Korean side to promote cooperation in Northeast Asia are suggested and discussed. The Chinese State Council amended the Environmental Protection Provisions and Clean Air Act and announced The Action Plan for Air Pollution Control (2013-2017), focusing on three key regions, Beijing-Tianjin-Hebei area (Jing-Jin-Ji), Yangtze River Delta (YRD) and Pearl River Delta (PRD). These policy actions and plan are mainly for the reducing coal usage and emissions from vehicles. It is suggested that, Korea should actively promote multi-national cooperation in the region to take an initiative role in environmental areas.