• Journal of Intelligence and Information Systems
• /
• v.27 no.3
• /
• pp.175-197
• /
• 2021

Recently, with the development of deep learning technology, research on unstructured data analysis is being actively conducted, and it is showing remarkable results in various fields such as classification, summary, and generation. Among various text analysis fields, text classification is the most widely used technology in academia and industry. Text classification includes binary class classification with one label among two classes, multi-class classification with one label among several classes, and multi-label classification with multiple labels among several classes. In particular, multi-label classification requires a different training method from binary class classification and multi-class classification because of the characteristic of having multiple labels. In addition, since the number of labels to be predicted increases as the number of labels and classes increases, there is a limitation in that performance improvement is difficult due to an increase in prediction difficulty. To overcome these limitations, (i) compressing the initially given high-dimensional label space into a low-dimensional latent label space, (ii) after performing training to predict the compressed label, (iii) restoring the predicted label to the high-dimensional original label space, research on label embedding is being actively conducted. Typical label embedding techniques include Principal Label Space Transformation (PLST), Multi-Label Classification via Boolean Matrix Decomposition (MLC-BMaD), and Bayesian Multi-Label Compressed Sensing (BML-CS). However, since these techniques consider only the linear relationship between labels or compress the labels by random transformation, it is difficult to understand the non-linear relationship between labels, so there is a limitation in that it is not possible to create a latent label space sufficiently containing the information of the original label. Recently, there have been increasing attempts to improve performance by applying deep learning technology to label embedding. Label embedding using an autoencoder, a deep learning model that is effective for data compression and restoration, is representative. However, the traditional autoencoder-based label embedding has a limitation in that a large amount of information loss occurs when compressing a high-dimensional label space having a myriad of classes into a low-dimensional latent label space. This can be found in the gradient loss problem that occurs in the backpropagation process of learning. To solve this problem, skip connection was devised, and by adding the input of the layer to the output to prevent gradient loss during backpropagation, efficient learning is possible even when the layer is deep. Skip connection is mainly used for image feature extraction in convolutional neural networks, but studies using skip connection in autoencoder or label embedding process are still lacking. Therefore, in this study, we propose an autoencoder-based label embedding methodology in which skip connections are added to each of the encoder and decoder to form a low-dimensional latent label space that reflects the information of the high-dimensional label space well. In addition, the proposed methodology was applied to actual paper keywords to derive the high-dimensional keyword label space and the low-dimensional latent label space. Using this, we conducted an experiment to predict the compressed keyword vector existing in the latent label space from the paper abstract and to evaluate the multi-label classification by restoring the predicted keyword vector back to the original label space. As a result, the accuracy, precision, recall, and F1 score used as performance indicators showed far superior performance in multi-label classification based on the proposed methodology compared to traditional multi-label classification methods. This can be seen that the low-dimensional latent label space derived through the proposed methodology well reflected the information of the high-dimensional label space, which ultimately led to the improvement of the performance of the multi-label classification itself. In addition, the utility of the proposed methodology was identified by comparing the performance of the proposed methodology according to the domain characteristics and the number of dimensions of the latent label space.

• Journal of the Korean Applied Science and Technology
• /
• v.36 no.4
• /
• pp.1312-1326
• /
• 2019

The purpose of this study is to examine whether the 12-week Nordic walking can improve the physical function and arthritis pain of elderly women with osteoarthritis This study were divided into randomly assigned Nordic Walking Exercise Group (n=9) and Control Group (n=7) for 16 Elderly women diagnosed with Osteoarthritis (age: 73±3.79 year, height: 154.3±4.09 cm). The exercise group used Nordic sticks to carry out 30 minutes of Nordic walking exercise three times a week for 12 weeks, and the kinetic intensity was set at 40-60% of HRR. The control group maintained daily life for the same period. Body composition (weight, percentage body fat, skeletal muscle mass), muscular strength, Flexibility (muscular strength of upper and lower limbs, flexibility of upper and lower limbs), balance ability (static balance, dynamic balance) and pain level were measured as subordinate variables. These indicators were measured twice before and after the exercise program. The study shows that percentage body fat and skeletal muscle mass in the body composition function over 12 weeks of Nordic walking exercise have significant effects after the exercise than before (p=004)(p=.003), and it also shows significant interaction effects between the groups and timings(p=.018)(p=.005). In muscular strength, Flexibility factors, there were significant effects between the groups and timings in the upper limb muscular strength and the lower limb flexibility (p=.009)(p=.036), and a significant difference between the exercise group and the control group(p=.006) in the lower limb muscular strength. In addition, in the upper limb flexibility, there was a more significant difference after the exercise than before(p=.020). There were improvement effects after the exercise than before in the balance ability and the static balance(p=.016), but no difference in the dynamic balance(p>.05). In pain, there was a significant improvement after the exercise than before(p=.022), and a significant difference between the exercise group and the control group(p=.013). In conclusion, the 12-week Nordic walking exercise has positive effects on the body composition functions of the elderly women with Osteoarthritis, and has a positive effect on the improvement of upper limb muscular strength and lower limb flexibility in the health fitness factors. These effects are believed to have contributed effectively to the improvement of the level of pain by contributing to the improvement of physical and motor functions of the elderly women with Osteoarthritis. Therefore, it is considered that Nordic walking exercise, which enhances stability and balance of the patients with Osteoarthritis by using poles, is an effective exercise method for the improvement of the body and motor functions by lowering the pain of the joints and reducing the muscular strength and percentage body fat.

• Journal of Intelligence and Information Systems
• /
• v.16 no.3
• /
• pp.163-179
• /
• 2010

Recentlythe concept of fast fashion is drawing attention as customer needs are diversified and supply lead time is getting shorter in fashion industry. It is emphasized as one of the critical success factors in the fashion industry how quickly and efficiently to satisfy the customer needs as the competition has intensified. Because the fast fashion is inherently susceptible to trend, it is very important for fashion retailers to make quick decisions regarding items to launch, quantity based on demand prediction, and the time to respond. Also the planning decisions must be executed through the business processes of procurement, production, and logistics in real time. In order to adapt to this trend, the fashion industry urgently needs supports from intelligent quick response(QR) system. However, the traditional functions of QR systems have not been able to completely satisfy such demands of the fast fashion industry. This paper proposes an intelligent quick response system for the fast fashion(IQRS-FF). Presented are models for QR process, QR principles and execution, and QR quantity and timing computation. IQRS-FF models support the decision makers by providing useful information with automated and rule-based algorithms. If the predefined conditions of a rule are satisfied, the actions defined in the rule are automatically taken or informed to the decision makers. In IQRS-FF, QRdecisions are made in two stages: pre-season and in-season. In pre-season, firstly master demand prediction is performed based on the macro level analysis such as local and global economy, fashion trends and competitors. The prediction proceeds to the master production and procurement planning. Checking availability and delivery of materials for production, decision makers must make reservations or request procurements. For the outsourcing materials, they must check the availability and capacity of partners. By the master plans, the performance of the QR during the in-season is greatly enhanced and the decision to select the QR items is made fully considering the availability of materials in warehouse as well as partners' capacity. During in-season, the decision makers must find the right time to QR as the actual sales occur in stores. Then they are to decide items to QRbased not only on the qualitative criteria such as opinions from sales persons but also on the quantitative criteria such as sales volume, the recent sales trend, inventory level, the remaining period, the forecast for the remaining period, and competitors' performance. To calculate QR quantity in IQRS-FF, two calculation methods are designed: QR Index based calculation and attribute similarity based calculation using demographic cluster. In the early period of a new season, the attribute similarity based QR amount calculation is better used because there are not enough historical sales data. By analyzing sales trends of the categories or items that have similar attributes, QR quantity can be computed. On the other hand, in case of having enough information to analyze the sales trends or forecasting, the QR Index based calculation method can be used. Having defined the models for decision making for QR, we design KPIs(Key Performance Indicators) to test the reliability of the models in critical decision makings: the difference of sales volumebetween QR items and non-QR items; the accuracy rate of QR the lead-time spent on QR decision-making. To verify the effectiveness and practicality of the proposed models, a case study has been performed for a representative fashion company which recently developed and launched the IQRS-FF. The case study shows that the average sales rateof QR items increased by 15%, the differences in sales rate between QR items and non-QR items increased by 10%, the QR accuracy was 70%, the lead time for QR dramatically decreased from 120 hours to 8 hours.

• Food Science and Preservation
• /
• v.20 no.3
• /
• pp.323-329
• /
• 2013

Strawberries are among the leading ready-to-eat agricultural products that have superior taste and nutrition. Thus, consumer concerns about the safety of eating strawberries are growing. To evaluate the contamination levels of strawberries according to their cultivation methods (nutriculture, pesticide-free culture and organic farming) and parts [fruit (flesh), stalk (pedunle) and leaf (calyx)], 1,020 parts of strawberry samples were collected at 12 farms in Nonsan-si and quantitatively or qualitatively examined for the indicators of food safety and food poisoning bacteria. The total aerobic bacteria count in the whole samples was 2.3~6.8 ${\log}_{10}$ CFU/g, and coliform bacteria were detected in 14.2% of the whole samples with a contamination level range of 2.1~4.5 log CFU/g. E. coli were detected in 0.9% of the whole samples with a contamination level range of 2.1~2.8 log CFU/g. The analysis of the bacterial levels according to the cultivation methods showed that the total aerobic bacteria and coliform counts were higher in the strawberries that were grown via organic farming than in those that were grown via nutriculture and pesticide-free culture. However, the E. coli counts of the strawberries that were grown via organic farming and via pesticide-free culture were similar and differed from that of the strawberries that were grown via nutriculture. The analysis of the contamination levels according to the parts of the strawberries showed that the total aerobic bacteria, coliform and E. coli counts of the fruits, stalks and leaves of the strawberries did not significantly differ. Staphylococcus aureus was detected in two organically grown strawberries, but Salmonella spp., Listeria monocytogenes and E. coli O157:H7 were not detected in the whole samples. These results show that the bacterial contamination levels of the strawberries differed based on their cultivation methods. Thus, a suitable method of reducing the bacterial contamination levels of strawberries according to their farming methods is needed.

• Journal of Intelligence and Information Systems
• /
• v.17 no.4
• /
• pp.241-254
• /
• 2011

Financial time-series forecasting is one of the most important issues because it is essential for the risk management of financial institutions. Therefore, researchers have tried to forecast financial time-series using various data mining techniques such as regression, artificial neural networks, decision trees, k-nearest neighbor etc. Recently, support vector machines (SVMs) are popularly applied to this research area because they have advantages that they don't require huge training data and have low possibility of overfitting. However, a user must determine several design factors by heuristics in order to use SVM. For example, the selection of appropriate kernel function and its parameters and proper feature subset selection are major design factors of SVM. Other than these factors, the proper selection of instance subset may also improve the forecasting performance of SVM by eliminating irrelevant and distorting training instances. Nonetheless, there have been few studies that have applied instance selection to SVM, especially in the domain of stock market prediction. Instance selection tries to choose proper instance subsets from original training data. It may be considered as a method of knowledge refinement and it maintains the instance-base. This study proposes the novel instance selection algorithm for SVMs. The proposed technique in this study uses genetic algorithm (GA) to optimize instance selection process with parameter optimization simultaneously. We call the model as ISVM (SVM with Instance selection) in this study. Experiments on stock market data are implemented using ISVM. In this study, the GA searches for optimal or near-optimal values of kernel parameters and relevant instances for SVMs. This study needs two sets of parameters in chromosomes in GA setting : The codes for kernel parameters and for instance selection. For the controlling parameters of the GA search, the population size is set at 50 organisms and the value of the crossover rate is set at 0.7 while the mutation rate is 0.1. As the stopping condition, 50 generations are permitted. The application data used in this study consists of technical indicators and the direction of change in the daily Korea stock price index (KOSPI). The total number of samples is 2218 trading days. We separate the whole data into three subsets as training, test, hold-out data set. The number of data in each subset is 1056, 581, 581 respectively. This study compares ISVM to several comparative models including logistic regression (logit), backpropagation neural networks (ANN), nearest neighbor (1-NN), conventional SVM (SVM) and SVM with the optimized parameters (PSVM). In especial, PSVM uses optimized kernel parameters by the genetic algorithm. The experimental results show that ISVM outperforms 1-NN by 15.32%, ANN by 6.89%, Logit and SVM by 5.34%, and PSVM by 4.82% for the holdout data. For ISVM, only 556 data from 1056 original training data are used to produce the result. In addition, the two-sample test for proportions is used to examine whether ISVM significantly outperforms other comparative models. The results indicate that ISVM outperforms ANN and 1-NN at the 1% statistical significance level. In addition, ISVM performs better than Logit, SVM and PSVM at the 5% statistical significance level.

• Korean Journal of Biological Psychiatry
• /
• v.9 no.2
• /
• pp.140-151
• /
• 2002

Object and Method:Minor physical anomalies(MPAs) are frequently seen in patients with schizophrenia. MPAs are considered to arise from the anomalous development of ectoderm-originated tissues in the developing fetus. Since the central nervous system originates from ectoderm, MPAs can be regarded as externally observable and objective indicators of the aberrant development which might have taken place in the central nervous system. To investigate whether MPAs are more frequent in schizophrenic patients, the frequencies of MPAs were compared between schizophrenic patients and normal controls. Total 245 schizophrenic patients diagnosed with DSM-IV(male : 158, female : 87), and 418 normal control subjects(male : 216, female : 202) were included in this study. The MPAs were measured using the modified Waldrop scale with fifteen items in six bodily regions; head, eye, ear, mouth, hand, and foot. Result:The total scores of Waldrop scale were $4.40{\pm}1.93$($mean{\pm}standard$ deviation) in patients and $3.43{\pm}1.68$ in controls for females, and for males, $4.58{\pm}1.75$ in patients and $4.28{\pm}1.59$ in controls. For females, the excess of MPAs in schizophrenic patients was statistically significant(t-test : p<0.001). For males, schizophrenic patients also showed more MPAs than normal controls, but this tendency did not reach statistical significance (t-test : p=0.094). When the modified Waldrop total scores excluding head circumference were compared, the total scores in schizophrenic patients were significantly higher for both male and female subjects(t-test : male p<0.001, female p=0.001). The individual anomaly items included in Waldrop scale were also investigated. The items of epicanthus, hypertelorism, malformed ears, syndactylia were significantly more frequent in schizophrenic patients. In contrast, the items of adherent ear lobes, asymmetric ears, furrowed tongue, curved fifth finger, single palmar crease and big gap between toes did not show any differences in frequency between schizophrenic patients and normal controls. Since a lot of statistical analyses showed different results between male and female subjects, it seems to be necessary to consider gender as an important controlling variable for the analysis, however only the item of head circumference showed statistically significant gender-related difference according to log-linear analysis. Conclusion:With a relatively large sample size, the frequencies of MPAs enlisted in Waldrop scale were compared between schizophrenic patients and normal controls in this study. MPAs were more frequently seen in schizophrenic patients and, especially, several specific items in the Waldrop scale showed prominent excess in schizophrenic patients. Although definite conclusions cannot be drawn due to the inherent limitation of the study using Waldrop scale, these results seem to support the possibility that aberrant neurodevelopmental process might be involved in the pathogenesis of schizophrenia in some of the patients.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.25 no.1
• /
• pp.101-119
• /
• 2022

Existing algal bloom monitoring is based on field sampling, and there is a limit to understanding the spatial distribution of algal blooms, such as the occurrence and spread of algae, due to local investigations. In this study, algal bloom monitoring was performed using an unmanned aerial vehicle and multispectral sensor, and data on the distribution of algae were provided. For the algal bloom monitoring site, data were acquired from the Mulgeum·Mae-ri site located in the lower part of the Nakdong River, which is the areas with frequent algal bloom. The Chlorophyll-a(Chl-a) value of field-collected samples and the Chl-a estimation formula derived from the correlation between the spectral indices were comparatively analyzed. As a result, among the spectral indices, Maximum Chlorophyll Index (MCI) showed the highest statistical significance(R²=0.91, RMSE=8.1mg/m³). As a result of mapping the distribution of algae by applying MCI to the image of August 05, 2021 with the highest Chl-a concentration, the river area was 1.7km², the Warning area among the indicators of the algal bloom warning system was 1.03km²(60.56%) and the Algal Bloom area occupied 0.67km²(39.43%). In addition, as a result of calculating the number of occurrence days in the area corresponding to the "Warning" in the images during the study period (July 01, 2021~November 01, 2021), the Chl-a concentration above the "Warning" level was observed in the entire river section from 12 to 19 times. The algal bloom monitoring method proposed in this study can supplement the limitations of the existing algal bloom warning system and can be used to provide information on a point-by-point basis as well as information on a spatial range of the algal bloom warning area.

• The Journal of Korean Society for Radiation Therapy
• /
• v.32
• /
• pp.61-71
• /
• 2020

Purpose: To investigate the effect of collimator angle on plan quality of PAN-Pelvis Multi-isocenter VMAT plan, dose reproducibility at the junction and impact on set-up error at the junction. Material and method: 10 adult patients with whole pelvis cancer including PAN were selected for the study. Using Trubeam STx equipped with HD MLC, we changed the collimator angle to 20°, 30°, and 45° except 10° which was the default collimator angle in the Eclipse(version 13.7) and all other treatment conditions were set to be the same for each patient and four plans were established also. To evaluate these plans, PTV coverage, coverage index(CVI) and homogeneity index (HI) were compared and clinical indicators for each treatment sites in normal tissues were analyzed. To evaluate dose reproducibility at the junction, the absolute dose was measured using a Falmer type ionization chamber and dose changes at the junction were evaluated by moving the position of the isocenter in and out 1~3mm and setting up the virtual volume at the junction. Result: CVI mean value was PTV-45 0.985±0.004, PTV-55 0.998±0.003 at 45° and HI mean value was PTV-45 1.140±0.074, and PTV-55 1.031±0.074 at 45° which were closest to 1. V20Gy of the kidneys decreased by 9.66% and average dose of bladder and V30 decreased by 1.88% and 2.16% at 45° compared to 10° for the critical organs. The dose value at the junction of the plan and the actual measured were within 0.3% and within tolerance. At the junction, due to set-up error the maximum dose increased to 14.56%, 9.88%, 8.03%, and 7.05%, at 10°, 20°, 30°, 45°, and the minimum dose decreased to 13.18%, 10.91%, 8.42%, and 4.53%, at 10°, 20°, 30°, 45° Conclusion: In terms of CVI, HI of PTV and critical organ protection, overall improved values were shown as the collimator angle increased. The impact on set-up error at the junction by collimator angle decreased as the angle increased and it will help improve the anxiety about the set up error. In conclusion, the collimator angle should be recognized as a factor that can affect the quality of the multi-isocenter VMAT plan and the dose at the junction, and be careful in setting the collimator angle in the treatment plan.

• Journal of agricultural medicine and community health
• /
• v.15 no.2
• /
• pp.97-106
• /
• 1990

1. Background Women's health and their involvement in health care are essential to health for everyone. If they are ignorant, malnourished or over-worked, the health &-their families as well as their own health will suffer. Women's health depends on broad considerations beyond medicine. Among other things, it depends upon their work in farming. their subordination to their families, their accepted roles, and poor hygiene with poorly equipped housing and environmental sanitation. 2. Objectives and Contents a. The health status of rural women : physical and mental complaints, experience of pesticides intoxication, Farmer's syndrome, experiences of reproductive health problems. b. participation in and attitudes towards housework and farming c. accessibility of medical care d. status of maternal health : fertility, family planning practice. induced abortion, and maternal care 3. Research method A nationwide field survey, based on stratified random sampling, was conducted during July, 1986. Revised Cornell Medical index(68 out of 195 items). Kawagai's Farmers Syndrome Scale, and self-developed structured questionnaires were used to rural farming wives(n=2.028). aged between 26-55. 4. Characteristics of the respondents mean age : 40.2 marital status : 90.8% married mean no. of household : 4.9 average years of education : 4.7 yrs. average income of household : \235,000 average years of residence in rural area : 36.4 yrs average Working hours(household and farming) : 11 hrs. 23 min 5. Health Status of rural women a. The average number of physical and mental symptoms were 12.4, 4.7, and the rate of complaints were 22.1%, 38.8% each. revealing complaints of mental symptomes higher than physical ones. b. 65.4% of rural women complained of more than 4 symptoms out of 9, indicating farmer's syndrome. 11.9 % experienced pesticide overdue syndrome c. 57.6% of respondents experienced women-specific health problems. d. Age and education of respondents were the variables which affect on the level of their health 6. Utilization of medical services a. The number of symptoms and complaints of respondents were dependent on the distance to where the health-care service is given b. Drug store was the most commonly utilized due to low price and the distance to reach. while nurse practitioners were well utilized when there were nurse practitioner's office in their villages. c. Rural women were internalized their subordination to husbands and children, revealing they are positive(93%) in health-care demand for-them but negative(30%) for themselves d. 33.0% of respondents were habitual drug users, 4.5% were smokers and 32.3% were alcohol drinkers. and 86.3% experienced induced-abortion. But most of them(77.6%) knew that those had negative effects on health. 7. Maternal Health Care a. Practice rate of contraception was 48.1% : female users were 90.9% in permanent and 89.6% in temporary contraception b. Induced abortions were taken mostly at hospital(86.3%), while health centers(4.7%), midwiferies(4.3%). and others(4.5%) including drug stores were listed a few. The repeated numbers of induced abortion seemed affected on the increasing numbers of symptoms and complaints. c. The first pre-natal check-up during first trimester was 41.8%, safe delivery rate was 15.6%, post-natal check-up during two months after delivery. Rural women had no enough rest after delivery revealing average days of rest from home work and farming 8.3 and 17.2. d. 86.6% practised breast feeding, showing younger and more educated mothers depending on artificial milk 8. Recommendations a. To lessen the multiple role over burden housing and sanitary conditions should be improved, and are needed farming machiner es for women and training on the use of them b. Health education should begin at primary school including health behavior and living environment. c. Women should be encouraged to become policy-makers as well as administrators in the field of women specific health affairs. d. Women's health indicators should be developed and women's health surveillance system too.

• Korean Journal of Environmental Biology
• /
• v.41 no.1
• /
• pp.18-30
• /
• 2023

Forest destruction is an inevitable result of the development processes. According to the environmental impact assessment, over 10% of the destroyed trees need to be recycled and transplanted to minimize the impact of forest destruction. However, the rate of successful transplantation is low, leading to a high rate of tree death. This is attributable to a lack of consideration for environmental factors when choosing a temporary site for transplantation and inadequate management. To monitor transplanted trees, a field survey is essential; however, the spatio-temporal aspect is limited. This study evaluated the applicability of remote sensing for the effective monitoring of transplanted trees. Vegetation indices based on satellite remote sensing were derived to detect time-series changes in the status of the transplanted trees at three temporary transplantation sites. The mortality rate and vitality of transplanted trees before and after the transplant have a similar tendency to the changes in the vegetation indicators. The findings of this study showed that vegetation indices increased after transplantation of trees and decreased as the death rate increased and vitality decreased over time. This study presents a method for assessing newly transplanted trees using satellite images. The approach of utilizing satellite photos and the vegetation index is expected to detect changes in trees that have been transplanted across the country and help to manage tree transplantation for the environmental impact assessment.