• Journal of Wetlands Research
• /
• v.26 no.2
• /
• pp.147-159
• /
• 2024

Lack of streamflow observations makes model calibration difficult and limits model performance improvement. Satellite-based remote sensing products offer a new alternative as they can be actively utilized to obtain hydrological data. Recently, several studies have shown that artificial intelligence-based solutions are more appropriate than traditional conceptual and physical models. In this study, a data-driven approach combining various recurrent neural networks and decision tree-based algorithms is proposed, and the utilization of satellite remote sensing information for AI training is investigated. The satellite imagery used in this study is from MODIS and SMAP. The proposed approach is validated using publicly available data from 25 watersheds. Inspired by the traditional regionalization approach, a strategy is adopted to learn one data-driven model by integrating data from all basins, and the potential of the proposed approach is evaluated by using a leave-one-out cross-validation regionalization setting to predict streamflow from different basins with one model. The GRU + Light GBM model was found to be a suitable model combination for target basins and showed good streamflow prediction performance in ungauged basins (The average model efficiency coefficient for predicting daily streamflow in 25 ungauged basins is 0.7187) except for the period when streamflow is very small. The influence of satellite remote sensing information was found to be up to 10%, with the additional application of satellite information having a greater impact on streamflow prediction during low or dry seasons than during wet or normal seasons.

• Journal of the Korean Institute of Landscape Architecture
• /
• v.52 no.1
• /
• pp.71-86
• /
• 2024

Despite the increasing importance of urban parks' ecological functions in dealing with the climate crisis, ecological parks are not clearly defined in Korea's legal system. Numerous ecological parks created nationwide cannot be systematically designated and managed due to various legal bases and varying management authorities. It is important to clarify the legal status of ecological parks in order to lead the ecological paradigm shift of urban parks and to improve the natural park system for a comprehensive and integrated approach to protect the national ecosystem. To this end, related laws were analyzed to identify problems and to draw directions for legislative improvement. Through the literature review of relevant laws, acts, and ordinances, six major directions for improvement were suggested based on the analysis of problems. First, the legal status of ecological parks in the administrative dichotomy of the current park system is ambiguous, and ecological parks should be clarified through the revision of park-related laws. Second, an ecological park can be defined as a sustainable park created and managed in an ecological manner, promoting the protection and restoration of the ecosystem, conservation, and promotion of biodiversity, and balancing nature observation, ecological learning, and leisure activities. Third, the role of the state and local governments should be systematically revised to lead to a new park planning and management model through new governance. Fourth, since the characteristics of ecological parks are affected by individual laws, the possibility of overlapping ecological parks for other uses should be allowed. Fifth, detailed guidelines and standard ordinances need to be enacted to meet the goals, principles, and facilities of ecological parks. Lastly, along with the revision of the laws, ordinances by local governments also need to be more concrete. This study, which tracks various legal realities related to ecological parks, can contribute to policymaking that can systematize the foundation for the creation of ecological parks to preserve nationwide ecosystems and provide citizens with opportunities to experience and learn about nature.

• The Sea:JOURNAL OF THE KOREAN SOCIETY OF OCEANOGRAPHY
• /
• v.13 no.2
• /
• pp.89-105
• /
• 2008

Sea level and atmospheric pressure data of 1999-2005 from four stations along the Korean east coast were analyzed to understand the sea level variability and its causal mechanism. The results of the wavelet and the auto-spectrum analyses indicate that the sea level fluctuations of 3-17 day period are statistically significant at the 95% confidence level, especially in spring to early summer. In this period, the coherency between the sea levels and the atmospheric pressures in a cross-spectrum is high, implying the importance of an inverted barometric effect in generation of the sea level fluctuations. To learn about the sea level variability, the cross-spectrum analyses were applied between the sea levels of the adjacent stations. The results show a case of southward phase propagations along the coast, as in 1999, 2003 and 2005, and an another case of no progressive phase lags between the stations, as in 2000-2002, and 2004. The phase speed in the former case is 12-15 m/s, which is a commonly observed phase speed of coastal Kelvin waves. Generation of such fluctuations seems to be related to low pressure cells developed in the Asian continent in spring and summer and moving eastward over the coastal region north of the stations. The latter case of no progressive phase lag, however, occurs when the low pressure cells developed in the continent move along the region south of the stations. In this case, the northeastward phase propagation with a speed of 5-8 m/s is observed along the southwestern coast of Japan.

• The Korean Society of Law and Medicine
• /
• v.24 no.4
• /
• pp.67-101
• /
• 2023

The utilization of generative AI in the medical field is also being rapidly researched. Access to vast data sets reduces the time and energy spent in selecting information. However, as the effort put into content creation decreases, there is a greater likelihood of associated issues arising. For example, with generative AI, users must discern the accuracy of results themselves, as these AIs learn from data within a set period and generate outcomes. While the answers may appear plausible, their sources are often unclear, making it challenging to determine their veracity. Additionally, the possibility of presenting results from a biased or distorted perspective cannot be discounted at present on ethical grounds. Despite these concerns, the field of generative AI is continually advancing, with an increasing number of users leveraging it in various sectors, including biomedical and life sciences. This raises important legal considerations regarding who bears responsibility and to what extent for any damages caused by these high-performance AI algorithms. A general overview of issues with generative AI includes those discussed above, but another perspective arises from its fundamental nature as a large-scale language model ('LLM') AI. There is a civil law concern regarding "the memorization of training data within artificial neural networks and its subsequent reproduction". Medical data, by nature, often reflects personal characteristics of patients, potentially leading to issues such as the regeneration of personal information. The extensive application of generative AI in scenarios beyond traditional AI brings forth the possibility of legal challenges that cannot be ignored. Upon examining the technical characteristics of generative AI and focusing on legal issues, especially concerning the protection of personal information, it's evident that current laws regarding personal information protection, particularly in the context of health and medical data utilization, are inadequate. These laws provide processes for anonymizing and de-identification, specific personal information but fall short when generative AI is applied as software in medical devices. To address the functionalities of generative AI in clinical software, a reevaluation and adjustment of existing laws for the protection of personal information are imperative.

• Journal of Korean Ophthalmic Optics Society
• /
• v.18 no.1
• /
• pp.75-84
• /
• 2013

Purpose: Changes of refractive correction value in different age group were investigated. Regarding the inhibitive effects against myopia progression after wearing reverse geometry contact lenses and myopia control lenses (MC lenses), the effects after wearing single vision lenses were compared. Methods: We organized children between the ages of six and fifteen into three groups by age, and distributed fifty-seven reverse geometry contact lenses, fifty-six MC lenses and seventy-eight single vision lenses among them to be worn. Group 1 consisted of children aged ten and under, Group 2 consisted of children between the ages of eleven and fifteen, and Group 3 represents all of the study participants. The aim of this study was to learn the inhibitive effects against myopia progression attained by changes of refractive correction value and to verify their statistical significance at twelve months and under, thirteen to twenty-four months and twenty-five to thirty-six months. Results: Changes of refractive correction value by each length of use in Group 3 were as follows. For the age group of under twelve months, participants using the reverse geometric contact lens showed no change, while those using the MC or single vision lens had significant changes (P<0.05) of $-0.36{\pm}0.10$ D and $-0.67{\pm}0.52$ D, respectively. Users of all three lens types displayed significant change (P<0.05), in the age group of between thirteen and twenty-four months, of $0.18{\pm}0.49$ D, $0.60{\pm}0.42$ D and $1.37{\pm}0.72$ D for users of the reverse geometry contact lens, the MC lens and the single vision lens, respectively. There were significant changes (P<0.05) of $0.29{\pm}0.61$ D, $0.93{\pm}0.57$ D and $1.72{\pm}0.78$ in the same respective order as the above in the age group of twenty-five to thirty-six months. Refractive correction value showed changes with different age group. Group 1 displayed significant changes (P<0.05) of $0.29{\pm}0.73$ D, $1.07{\pm}0.59$ D and $1.75{\pm}0.74$ D for users of the reverse geometry contact lens, MC lens and single vision lens, respectively, up to thirty-six months of lens wearing; Group 2, also up to thirty-six months, displayed significant changes (P<0.05) of $0.28{\pm}0.42$ D, $0.75{\pm}0.49$ D and $1.70{\pm}0.84$ D in the same respective order, and changes in refractive correction for the age group under ten years was significantly greater (P<0.05) for the age group of eleven and older. Conclusions: The results found in this study demonstrate that there were no changes of refractive correction value for the case of wearing reversing geometry contact lens up to twelve month or less. MC lens showed less changes in variations of visual acuity for all users which might be resulted in inhibiting progression of myoptia. When both reverse geometry contact lens and the MC lens are wearing for the period from 13 to 36 month, both lens showed less changes in variation of visual acuity for all users. The results suggested that the less changes in variation of visual acuity of both lens had an effect on inhibiting progression of myopia.

• Journal of Intelligence and Information Systems
• /
• v.23 no.1
• /
• pp.95-108
• /
• 2017

Recently, AlphaGo which is Bakuk (Go) artificial intelligence program by Google DeepMind, had a huge victory against Lee Sedol. Many people thought that machines would not be able to win a man in Go games because the number of paths to make a one move is more than the number of atoms in the universe unlike chess, but the result was the opposite to what people predicted. After the match, artificial intelligence technology was focused as a core technology of the fourth industrial revolution and attracted attentions from various application domains. Especially, deep learning technique have been attracted as a core artificial intelligence technology used in the AlphaGo algorithm. The deep learning technique is already being applied to many problems. Especially, it shows good performance in image recognition field. In addition, it shows good performance in high dimensional data area such as voice, image and natural language, which was difficult to get good performance using existing machine learning techniques. However, in contrast, it is difficult to find deep leaning researches on traditional business data and structured data analysis. In this study, we tried to find out whether the deep learning techniques have been studied so far can be used not only for the recognition of high dimensional data but also for the binary classification problem of traditional business data analysis such as customer churn analysis, marketing response prediction, and default prediction. And we compare the performance of the deep learning techniques with that of traditional artificial neural network models. The experimental data in the paper is the telemarketing response data of a bank in Portugal. It has input variables such as age, occupation, loan status, and the number of previous telemarketing and has a binary target variable that records whether the customer intends to open an account or not. In this study, to evaluate the possibility of utilization of deep learning algorithms and techniques in binary classification problem, we compared the performance of various models using CNN, LSTM algorithm and dropout, which are widely used algorithms and techniques in deep learning, with that of MLP models which is a traditional artificial neural network model. However, since all the network design alternatives can not be tested due to the nature of the artificial neural network, the experiment was conducted based on restricted settings on the number of hidden layers, the number of neurons in the hidden layer, the number of output data (filters), and the application conditions of the dropout technique. The F1 Score was used to evaluate the performance of models to show how well the models work to classify the interesting class instead of the overall accuracy. The detail methods for applying each deep learning technique in the experiment is as follows. The CNN algorithm is a method that reads adjacent values from a specific value and recognizes the features, but it does not matter how close the distance of each business data field is because each field is usually independent. In this experiment, we set the filter size of the CNN algorithm as the number of fields to learn the whole characteristics of the data at once, and added a hidden layer to make decision based on the additional features. For the model having two LSTM layers, the input direction of the second layer is put in reversed position with first layer in order to reduce the influence from the position of each field. In the case of the dropout technique, we set the neurons to disappear with a probability of 0.5 for each hidden layer. The experimental results show that the predicted model with the highest F1 score was the CNN model using the dropout technique, and the next best model was the MLP model with two hidden layers using the dropout technique. In this study, we were able to get some findings as the experiment had proceeded. First, models using dropout techniques have a slightly more conservative prediction than those without dropout techniques, and it generally shows better performance in classification. Second, CNN models show better classification performance than MLP models. This is interesting because it has shown good performance in binary classification problems which it rarely have been applied to, as well as in the fields where it's effectiveness has been proven. Third, the LSTM algorithm seems to be unsuitable for binary classification problems because the training time is too long compared to the performance improvement. From these results, we can confirm that some of the deep learning algorithms can be applied to solve business binary classification problems.

• The Korean Journal of Rehabilitation Nursing
• /
• v.2 no.2
• /
• pp.257-270
• /
• 1999

The purpose of this study is to investigate factors related to the intent of using home nursing of chronic disease patients who got out of a university hospital. For the purpose, the study selected 153 patients who were hospitalized and left K university hospital with diagnoses of cancer, hypertension, diabetes and cerebral vascular accident and ordered to be discharged and performed interviews with them and surveys on their medical records to obtain the following results. For this study a direct-interview survey and medical record review was conducted from June 28 to Aug. 30, 1998. The frequency and mean values were computed to find the characteristics of the study subjects, and $X^2$-test, t-test, factor analysis and multiple logistic regession analysis were applied for the analysis of the data. The following results were obtained. 1) When characteristics of the subjects were examined, men and women occupied for 58.8% and 41.2%, respectively. The subjects were 41.3 years old in aver age and had the monthly aver age earning of 0.99 million won or below, which was the most out of the total subjects at 34.6%. Among the total, 87.6% resided in cities and 12.4 in counties. The most left the hospital with diagnosis of cancer at 51.6%, followed by hyper tension at 24.2%, diabetes at 13.7% and cerebral vascular accident at 7.2%. 2) 93.5% of the selected patients had the intent of using home nursing and 6.5%, didn't. Among those patients having the intent, 85.6% had the intent of paying for home nursing and 14.4%, didn't. The subjects expected that the nursing would be paid 9,143 won in aver age and 47.7% of them preferred national authorities as the main servers. 86.3% of the subjects thought that home nursing business had the main advantage of making it possible to learn nursing methods at home and thereby contributing to improving the ability of patients and their facilities to solve health problems. 3) Relations between the intent of use and characteristics of the subjects such as demography-related social, home environment, disease and physical function characteristics did not show statistically significant differences among one another. Compared to those who had no intent of using home nursing, the group having the intent had more cases of male patients, the age of 39 or below, residence in cities, 5 family member s or more, no existence of home nursing servers, leaving the hospital from a non-hospitalized building, disease development for five months or below, hospitalization for ten days or more, non-hospitalization with in the recent one month, two times or over of hospitalization, leaving the hospital with no demand of special treatment, operation underwent, poor results of treatment, leaving the hospital with demand of rehabilitation services, physical disablement and high evaluation point of daily life. 4) Among those patients having the intent of using home nursing, 47.6% demanded technical nursing and 55.9%, supportive nursing. As technical nursing,' inject into a blood vessel ' and 'treat pustule and teach basic prevention methods occupied for 57.4%, respectively, topping the list. Among demands of supportive nursing, 'observe patients 'status and refer them to hospitals or community resources as available, if necessary' was the most with percent age point of 59.5. Regarding the intent of paying for home nursing, 39.2% of those patients wishing to use the nursing responded paying for technical services and 20.2, supportive services. In detail, 70.0% wanted to pay for a service stated as 'inject into a blood vessel', highest among the former services and 30.7%, a service referred to as 'teaching exercises needed to make the body of patients move', highest among the latter. When this was analyzed in terms of a relation between the need(the need for home nursing) and the demand(the intent of paying for home nursing), The rate of the need to the demand was found two or three times higher in technical nursing(0.82) than in supportive nursing(0.35). In aspects of tech ical nursing, muscle injection(1.26, the 1st rank) was highest in the rate while among aspects of supportive nursing, a service referred to as 'teach exercises needed for making patients move their bodies normally'(0.58, the 1st rank). 5) factors I(satisfaction with hospital services), II(recognition of disease state), III(economy) and IV(period of disease) occupied for 34.4, 13.8, 11.9 and 9.2 percents, respectively among factors related to the intent by the subjects of using home nursing, totaled 59.3%. In conclusion, most of chronic disease patients have the intent of using hospital-based home nursing and satisfaction with hospital services is a factor affecting the intent most. Thus a post-management system is needed to continue providing health management to those patients after they leave the hospital. Further, supportive services should be provided in order that those who are satisfied with hospital services return to their community and live their in dependent lives. Based on these results, the researcher would make the following recommendation. 1) Because home nursing becomes more and more needed due to a sharp increase in chronic disease patients and elderly people, related rules and regulations should be made and implemented. 2) Hospital nurses specializing in home nursing should be cultivated.

• Journal of Environmental Health Sciences
• /
• v.17 no.2
• /
• pp.54-66
• /
• 1991

As a result of measuring illumination and making up a question at home visit directly by investigator who trained over twenty days period from October 4 to 24, 1990, in order to render help which illumination problem against house, society against eyes or framing of health instruction potgram by seizing natural lighting actual conditions of house and actual conditions of wearing spectacles and by investigating interrelationship, I can summarize as follows. 1) In property of investigation subject, woman 66.9%, In an age, the twenties was largest of 27.4%, the forties was 20.2%, the fifties was 18.6%, the thirties was 17.4%. In academic career, those of upper secondary school grauates was largest of 28.6%, those who possess university career was 25.9%, those who middle school career was 20.9%, decoding of Korean alphabet was 2%. 2) By a residence area, a big city was 43.3%, farming and fishing villages were 20.3%, the rest was a small town and the administrative office of town, township. In positon of house, the middle area was 43.6%, resident of suburb area was 38.0%. In form of house, a Korean-style house was 40.8%, a western-style house was 34.8%, an apartment house was 11.0%. In the a standard of living, the middle classes 77.2%, the lower classes were 15.3%. In residential house unit of area, from 21 to 30 unit of area was largest of 31.5%, from 10 to 20 unit of area was 19.9%, from 31 to 40 was 18.7%. 3) The wearing spectacles rate of study user was 44.1%. By the area, those who wearing spectacles was more than a half of 50.8% in the resident of big city area. As passing from the farm area to the city, that is being resident of big city was high wearing spectacles rate. In position of house, as being residence in central street showed high wearing spectacles rate. (central street was 51.5%, the middle area was 44.5% and the suburb area was 40.1%.) It seemed similarity difference a variable by position of house from wearing spectacles in standard of 1%. By form of house, wearing spectacles rate those who resident in apartment house was 49.5%, that rate those who resident in a western-style house was high of 49.0%, that rate those who resident in a Korean-style house was the lowest 39.0%. By social position of resident in room, in students case who study showed very high, as university students were very high of 62.3% idn wearing spectacles rate, middle and high school students 'were 50.0%, members of society were 47.6%, workers 20.3%. It seemed similarity difference from academic career in standard of 1%. By an age, the thirties was high of 54.1% in wearing spectacles rate, the twenties was 43.2%, the teenage was the lowest of 11.8%. 4) In illumination of study, over 200Lux was high of 40.1%. but below 99Lux which inappropriate illumination to see the books was 32.4%. Average by area, below 99Lux was 22.7% and over 400Lux was 50.0% in case of wooden floor. As examine by area, below 99Lux was high of 27.0% a case of wooden floor in the big city area, it was not good in illumination passing from the farm area(15.0%) to the city(19.0%). Average illlumination by area of the main living room below 99Lux was high of 37.5%, less than 200Lux was 58.5% of whole. In general, illumination of the main livingroom was inappropriate. By area, the big city was 32.5% below 99Lux, the middle and small city area were 33.8%, town and township area were 45.0%, farming and fishing area were 42.8%. By area, in the big city, illumination of study was 52.5% over 200Lux and 28.9% below 99Lux. In case of the middle and small city, study user of below 99Lux was 38.8% and over 200Lux was 46.9%. In case of the seat of town township, below 99Lux was 34.1% and over 200Lux was 39.7%. In case of farming and fishing area, illumination of study was 33.4% below 99Lux and 48.4% over 200Lux. It tends to high rate of inappropriate illumination. 5) By position of house, in case of wooden floor, less than 100Lux was 24.5% in central street. It was bad illumination than others position of house. In case of the main livingroom, less than 100Lux was 40.4% in the suburb area. It was bad iliumnation than others position of house. In case of study, less than 100Lux was 35.4% in the middle area, it was worse in illumination. In case of the main living room, is seemed similarity difference in standard of 1%. 6) By form of house, in case of wooden floor, illumination of less than 100Lux was 23.8% in a western-style house, it was bad illumination than others form of house. In case of the main livingroom, illumination of less than 100Lux was 47.4% in a Korean-style house, it was remarkably bad illumination than others form of house. In case of study, a Korean-style house was 38.8%, it was very bad illumination than others form of house. In case of the main livingroom and study, it seemed similatrity difference each as P < 0.01 and P < 0.05 in standard of 1%. 7) The wearing spectacles rate of those who use room of illumination over 400Lux was 40.7%, and that of those who use room of illumination less than 100Lux was 28.1%. It seemed similarity differecce in standard of 1%. 8) In period of wearing spectacles, 21.3% of total investigator-highest-was from before five years, 8.6% was from before three years. Among those who use of illumintion less than 99Lux, 34.0% began to wear spectacles from before two years 31.7% was from before five years, 30.3% was from before four years. It seemed similarity difference from period of wearing spectacles by illumination in standard of 1 %. 9) Among cause which sight grow worse, the first was that it was each 33.2% and 27.4% in response rate because watch TV nearly to wearing spectacles person and non-wearing person. The second was that a lot of seeing books was 25.3% in wearing spectacles person and response rate for dark illumination was 7.4% in nonwearing spectacles person. It seemed similarity difference in standard of 1%. (P < 0.01). 10) In experience which take medicine good for eyes, it was 50.1% in wearing spectacles person and 8.5% in non-wearing spectacles person. It seemed similarity difference in standard of 1%(P < 0.01). As we have seen above, inappropriate illumination can be a cause of wearing spectacles. Nevertheless, actually, is realities to indifferent against illumination of house. So it must learn knowledge about health obstacle of illumination through society instruction and school eduction against students as well as general residents. In case that natural lighting is inappropriate structural of house, we must be able to maintain appropriate illumination through artificial illumination. And so eyes which is core of human life have to be protected, related the authorities, related group, and all health medical personnel will organically cooperate with and make efforts.

• Journal of Music and Human Behavior
• /
• v.4 no.2
• /
• pp.66-83
• /
• 2007

The purpose of this study is to learn both effects of the vocal psychotherapy on the self-consciousness and the variety of the self-consciousness on the vocal psychotherapy in return. The research for this study was performed to three subjects who were students of E university, Seoul, ten times for sixty minutes. The subjects were all volunteers for the advertisement on a music-therapy program searching for them on the web site of E university. The vocal psychotherapy program consists of four steps and each of them consists of two to four short terms again. Both before and after the experiment, examinations on self-consciousness were done to recognize the change of the subjects' self-consciousness which would be caused by the vocal psychotherapy activity. After every short term, the subjects were asked to write reports to closely analyze the change of self-consciousness according to the terms and the variety of the subjects. The effect of the vocal psychotherapy activity on the changes of scores in the self-consciousness examination is the first thing to point out on this study. There appeared some personal varieties on the total scores of the examination and scores of some sub-categories. Especially, there were different scores on the private self-consciousness, the public self-consciousness, and the social anxiety between before and after performing the vocal psychotherapy program. Subject A, who had got the best score of all on the scope of the private self-consciousness, showed the steepest decrease on the very scope. On the contrary, the subject showed decrease of scores of the public self-consciousness and the social anxiety in the relatively little rate. Subject B, who had got the highest score of the three on the public self-consciousness, showed the steepest decrease on that of all scopes and showed no difference on the social anxiety scope. In the case of the last one, subject C, who had relatively low scores on the private and public self-consciousness than the others, the private self-consciousness score increased but the public self-consciousness and the social anxiety scores decreased. The changes of the scores of each questions were examined in order to see possible other changes that had not been exposed on the changes of the total and sub-categories scores. As a result of that, of all twenty-eight questions, there were changes about one to two points. Subject A showed the difference with thirteen questions, subject B with sixteen and subject C with nineteen questions. The rate of change of subject C was relatively small but more questions changed and the change of score was wider than the others. Considering all those results, It can be possibly said that the vocal psychotherapy affects the changes of the scores of sub-categories in self-consciousness examination. The next thing to point out on this study is the change of recognition that was exposed on the subjects' report after every short term of the program. As a result of the close analyzing, according to the short terms and variety of self-consciousness, recognizing the way express subjects themselves by voice and recognizing their own voices appeared to be different. How much they cared about others and why they did so were also different. According to the self reports, subject A cared much about her inner thought and emotion and tended to concentrate herself as a social object. There appeared some positive emotional experiments such as emotional abundance and art curiosities on her reports but at the same time some negative emotions such as state-trait anxiety and neuroticism also appeared. Subject B, who showed high scores on the private and public self-consciousness like subject A, had a similar tendency that concentrates on herself as a social object but she showed more social anxiety than subject A. Subject C got relatively lower points in self-consciousness examination, tended to care about herself, and had less negative emotions such as state-trait anxiety than other subjects. Also, with terms going on, she showed changes in the way of caring about her own voice and others. This study has some unique significances in helping people who have problems caused by self-estimation activated with self-consciousness, using voices closely related to one's own self, performing the vocal skills discipline to solve the technical problems. Also, this study has a potentiality that the vocal psychotherapy activity can be effectively used as a way affects the mental health and developing personality.

• 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.