• Korean Journal of Psychosomatic Medicine
• /
• v.17 no.2
• /
• pp.52-61
• /
• 2009

Objectives : The aim of this study was to assess health-related quality of life and evaluate the risk factors affecting severity of menopausal symptoms in early and late postmenopausal women based on the stages of reproductive aging workshop(STRAW) paradigm. Methods : This cross-sectional study examined 497 Korean postmenopausal women aged 41-59 years in Seoul and Gyeonggi province. We divided subjects into early postmenopause group and late postmenopause group. Menopause Rating Scale(MRS) was used to measure the quality of life. MRS scores, sociodemographic variables, smoking, alcohol use, age at menopause, and risk factors such as attitude to menopause, depression, history of premenstrual dysphoric disorder were compared between early and late postmenopause groups. Multiple regression analysis was performed in each group to assess the independent contribution of several variables. Results : Early postmenopause group showed significantly higher MRS scores, more negative attitude toward menopause, higher scores of depressive symptoms than late postmenopause group. Moderate to very severe hot flush group showed significantly, more negative attitude toward menopause, higher score of depressive symptoms, and higher MRS scores than none to mild hot flush groups. Depressive symptoms and attitude toward menopause contributed to the severity of menopausal symptom in both early and late postmenopause groups. Chronological age, age at menopause, history of PMDD contributed to severity of menopausal symptoms in early postmenopause group while marital status and occupation contributed in late postmenopause group. Conclusion : Health-related quality of life in postmenopause women was significantly lower in early postmenopause group than the late. Attitude toward menopause and depressive symptoms contributed significantly to quality of life in both early and late postmenopause groups but other variables contributed differently in each group. Further studies on clinical samples of postmenopausal women in order to confirm quality of life and its risk factor are needed to be done.

• Journal of Korean Society of Transportation
• /
• v.35 no.4
• /
• pp.261-277
• /
• 2017

This study defines influencing sections as the part of the road section where passing vehicles are traveling with the lower speed compared to speed limit due to speed humps. The influencing section was divided into 3 parts; influencing section before the speed hump, interval section, and influencing section after the speed hump. This analysis focused on the changes of each part depending on installation types, vehicle types, and daytime or nighttime. For the interval section, especially, the ratio of distance traveled with lower speed than speed limit to interval section is defined as effective influencing section ratio to be analyzed. Vehicle speed profiles were collected with a speed gun to extract influencing section lengths. The survival analysis was applied and estimated survival functions are compared with each other by several statistical tests. As a consequence, the average length of influencing section on the 50m sequential speed humps was 75.3% longer during the deceleration than that of isolated speed hump, and 18.9% during the acceleration. The effective influencing section ratio for the 30m and 50m sequential speed humps had a small difference of 81.0% and 76.0% while the absolute values of the section that passing speed were less than the speed limit were longer on 50m sequential speed humps, each being 24.3m and 38.0m. Using the log rank test, it was evident that sequential speed humps were more effective to increase the length of influencing sections compared to the isolated speed hump. Vehicle type was the strong factor for influencing section length on the isolated speed hump, but daytime or nighttime was not the effective one. This research result can be used for improving the efficiency selecting the installation point of speed humps for road safety and estimating the standard of the distance between sequential speed humps.

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

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

• Journal of the Korean Academy of Child and Adolescent Psychiatry
• /
• v.10 no.2
• /
• pp.201-211
• /
• 1999

Objectives：This study investigated the characteristics of detained delinquent adolescents and variables related to the repeated crime during 6 months after release. Methods：The socio-demographic and crime-related characteristics of 73 detained adolescents were evaluated by semi-structured interviews and police records, and the psychological characteristics of them measured by the MMPI. We also compared the characteristics between subjects with and without repeated crime during 6 months after release. Results：1) Most of detained adolescents had families with low socioeconomic status(77%) and broken families(48%). Sixty-six percent of them were dropped out of school. The most frequent crime pattern was theft(49%), and with accomplice(77%). Seventy-five percent of total subjects had the records of previous conviction. Of the previous convictions, seventy-eight percent was same with the present crimes. 2) Subjects with repeated crime during 6 months after release were younger and had higher T-score on Pa scale of MMPI than the subjects without repeated crime. More adolescents with repeated crime had broken families than those without repeated crime. They also showed the crime-related characteristics of higher percent of theft among crime patterns, higher incidence of previous conviction, younger age of the first crime, and shorter crime-free duration from the last to present crime. Conclusion：These results of present study suggest that the development and the persistence of adolescent delinquency would be resulted from interaction of factors of individual, family, school, and community. By the comparison between subjects with and without repeated crime, it was found that familial dysfunction, younger age at first crime, presence of previous conviction might be the risk factors for repeated delinquency. To prevent repeated crime of delinquent adolescents more effectively, early therapeutic intervention and the development of programs to help adaptation in school and community would be essential.

• Korean Journal of Psychosomatic Medicine
• /
• v.12 no.2
• /
• pp.157-164
• /
• 2004

Objectives : This study was designed to investigate depression, anxiety, alexithymia, stress responses between well-controlled and poorly-controlled diabetic patients by glycated hemoglobin levels. Methods : The subjects were 55 diabetic patients(mean age : $49.9{\pm}9.9$, 27 men and 28 women) who were confirmed to have diabetes depending on the laboratory findings as well as clinical symptoms at the St. Vincent Hospital Diabetes Clinic, from Mar. 2004 to Aug. 2004. Korean version of Beck Depression Inventory(BDI), State and Trait Anxiety Inventory(STAI), Toronto Alexithymia Scale(TAS) and Stress Response Inventory(SRI) were used for assessment. Based on glycated hemoglobin levels, the patients were divided into 10 well-controlled group(below 7%) and 45 poorly-controlled group(above 7%). We compared BDI, STAI, TAS and SRI scores between two groups by independent t-test. Results 1) Well-controlled diabetics, compared with poorly controlled group, manifested decreased illness duration($12.2{\pm}55.4$months vs. $55.4{\pm}66.6 months)(p=0.000), but other demographic data showed no difference between two groups. 2) The STAI scores of poorly-controlled group were significantly higher in both state anxiety sores $(38.7{\pm}3.8 \;vs.\;43.7{\pm}6.7)(p=0.29)$ and trait anxiety scores$(36.9{\pm}5.7\;vs.\;41.5{\pm}6.4)(p=0.43)$ than well-controlled groups. 3) No significant differences were found in the score of BDI, TAS, SRI between well and poorly-controlled diabetic groups. Conclusion : The above results suggest that poorly-controlled diabetic patients are more likely to have higher anxiety level than well-controlled diabetic patients. However, there were no differences in depression, alexithymia, stress responses between two group. We suggest that physicians should consider integrated approaches for psychiatric problems in the management of diabetic patients.