• Journal of Korea Multimedia Society
• /
• v.25 no.10
• /
• pp.1375-1385
• /
• 2022

In this paper, we propose a semi-supervised domain adaptation solution to deal with practical face recognition (FR) scenarios where a single face image for each target identity (to be recognized) is only available in the training phase. Main goal of the proposed method is to reduce the discrepancy between the target and the source domain face images, which ultimately improves FR performances. The proposed method is based on the Domain Adatation network (DAN) using an MMD loss function to reduce the discrepancy between domains. In order to train more effectively, we develop a novel loss function learning strategy in which MMD loss and cross-entropy loss functions are adopted by using different weights according to the progress of each epoch during the learning. The proposed weight adoptation focuses on the training of the source domain in the initial learning phase to learn facial feature information such as eyes, nose, and mouth. After the initial learning is completed, the resulting feature information is used to training a deep network using the target domain images. To evaluate the effectiveness of the proposed method, FR performances were evaluated with pretrained model trained only with CASIA-webface (source images) and fine-tuned model trained only with FERET's gallery (target images) under the same FR scenarios. The experimental results showed that the proposed semi-supervised domain adaptation can be improved by 24.78% compared to the pre-trained model and 28.42% compared to the fine-tuned model. In addition, the proposed method outperformed other state-of-the-arts domain adaptation approaches by 9.41%.

• Korean Journal of Applied Biomechanics
• /
• v.15 no.1
• /
• pp.91-109
• /
• 2005

The goal of this research was to determine whether gender differences exist in impact force and impact shock variables at stance phase during a preferred running. Ten male and ten female subjects volunteered to participate in this study. Impact force was quantified by using a surface-mounted force plate. In addition, Axial accelerations of the tibias and mouth were measured using low-mass accelerometers. Comparison of parameters relating to impact force and impact shock which attained from time domain, and impact shock parameters which were analyzed in frequency domain were made between genders. The conclusions based on results were as follows; 1. There were no significantly differences in impact force, mouth and tibia acceleration peak in time domain between two genders. 2. The male group was greater in impact shock peak of PSD(power spectral density) at the tibia than female group(p<.05), but no differences in active impact of PSD at the tibia and the mouth between two genders. 3. Female subjects exhibited that a peak of impact shock attenuation analyzed in frequency domain moved toward a high frequency, but no difference in time domain between two genders.

• Korean Journal of Cognitive Science
• /
• v.18 no.1
• /
• pp.1-14
• /
• 2007

In a goal-oriented dialogue, speakers' intentions can be represented by domain actions that consist of pairs of a speech act and a concept sequence. Therefore, if we plan to implement an intelligent dialogue system, it is very important to correctly infer the domain actions from surface utterances. In this paper, we propose a statistical model to determine speech acts and concept sequences using conditional random fields at the same time. To avoid biased learning problems, the proposed model uses low-level linguistic features such as lexicals and parts-of-speech. Then, it filters out uninformative features using the chi-square statistic. In the experiments in a schedule arrangement domain, the proposed system showed good performances (the precision of 93.0% on speech act classification and the precision of 90.2% on concept sequence classification).

• The Journal of Korean Physical Therapy
• /
• v.21 no.2
• /
• pp.55-63
• /
• 2009

Purpose: This study evaluated of quality of life (QOL) and physical therapy satisfaction (PTS) in caregivers of cerebral palsy children. Methods: Eighty six caregivers in Gwang-ju were examined. The caregivers' QOL was assessed through self-reports using the WHO Quality of Life-BREF (WHOQOL-BREF), and the data was analyzed separately for each of its 5 domains. Physical therapy satisfaction was assessed using a self-reported questionnaire and the data was analyzed separately for the 2 domains. The results were processed by the mean and standard deviation and then verified by ANOVA and a t-test to determine the significant differences in the QOL and PTS. The factors affecting the QOL were processed by logistic regression. Results: Life environment domain of the QOL across the caregivers factors showed a statistically significant difference in age at natal, monthly income and education. The birth order in the cerebral palsy child factor showed a significant difference in the general, physical and emotional domain of the QOL. The PTS showed a significant difference in the delivery of therapy domain and in the birth order and walking condition of the environment domain. Caregivers education was found to be associated independently with the life environment domain of QOL. Conclusion: Because the QOL of caregivers is an important treatment goal for children with cerebral palsy, early family intervention should be considered.

• Journal of Korean society of Dental Hygiene
• /
• v.23 no.5
• /
• pp.369-377
• /
• 2023

Objectives: The aim of this study was to develop and validate a Korean-adapted version of the job expectations scale for dental hygienists. Methods: To achieve this goal, a comprehensive methodology was employed, including frequency analysis, exploratory factor analysis (maximum likelihood, direct oblimin rotation), confirmatory factor analysis, Cronbach's α reliability, independent t-test, and one-way ANOVA (Duncan post-hoc test) were performed. Results: The Korean version of job expectations scale for dental hygienists comprised 21 items in the personal domain and 5 items in the social domain. The scale demonstrated high construct validity, convergent validity, discriminant validity, and reliability were found. While dental hygienists were found to have higher expectations in the their personal domain than the social domain, expectations in the social domain were higher among male hygienists, those with a bachelor's degree or higher, those in regular permanent employment, and those working in hospital-level or higher facilities. Conclusions: Job expectations is a more comprehensive construct than job roles or professional identity. We anticipate that the instrument will be widely used to analyze trends in job expectations in response to changes in societal values regarding to dental hygienists, the gap between realities and ideals within the dental hygiene profession, and policy considerations.

• Proceedings of the Korean Institute of Navigation and Port Research Conference
• /
• 1994.10a
• /
• pp.79-90
• /
• 1994

The goal of this research is to get a measuring system for absorption characteristics of electromagnetic wave absorbers in ferrite type which are lat grid-type special type etc. It is however very difficult to measure the absorption characteristics as in low frequency as in 30 MHz, To solve the problem therefore we propose a standing wave method and a measurement method in time domain using parallel striplines construct the measuring system and measure the characteristics of ferrite microwave absorbers in grid type using the proposed measuring system.

• Proceedings of the Korean Information Science Society Conference
• /
• 1999.10b
• /
• pp.101-103
• /
• 1999

Decentralized 정보는 여러 도메인에 대한 heterogeneous한 독립적인 정보가 자율적으로 존재하며 이들 정보간의 관계성의 고려한 전체에 대한 global view가 존재하지 않기 때문에 inter-domain에 대한 마이닝을 수행하는데 어려움이 있다. 본 연구에서는 intra-domain knowledge discovery, intra 및 inter-domain problem solving method라는 접근방법으로, decentralized 데이터 환경에서 문제 해결에 필요한 정보 추출을 위한 데이터 tailoring과 분산 데이터에 대한 목표-지향 데이터마이닝(goal-oriented data-mining)을 통해 문제 해결을 위해 필요한 지식을 생성하고 이들 간의 관련 정보를 탐색하여 문제를 해결하는 프레임워크를 제안한다. 특히, 생성된 지식간의 협동 문제 처리를 멀티 에이전트 패러다임을 이용하기로 한다. 제안 프레임워크는 산재되어 있는 데이터들로부터 문제 해결에 유용한 지식 차원의 정보를 추출해내고 생성된 지식을 바탕으로 각 도메인 정보에 대한 개별적인 사용뿐 만 아니라 서로 cooperation을 통한 문제 해결을 지원함으로써, 개방된 분산 환경하에 decentralized 되어 있는 여러 도메인 정보를 보다 효율적으로 활용할 수 있는 새로운 형태의 문제 해결 방법이라고 할 수 있다.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2001.10a
• /
• pp.291-296
• /
• 2001

The computerized automation of modern manufacturing system is tremendously important for the cost reduction. Recently, the automation is being realized by Computer Integrated Manufacturing (CIM) concept and the unmanned factory is the final goal of the automation. The automation of the low level of factory is the most crucial and difficult aspect in total CIM environment since various machines are involved. A cell is defined as a manufacturing group of closely related machineries in the entire factory hierarchy, The management scheme for a cell controller is defined from the software viewpoint. The domain of a cell is designed with tasks(processes) in a multiple-task operating system. The operation of the cell is established by defining schemes for overall control of the cell domain, databases, job scheduling, communications within the domain, user-interface and communications with outside machines. Although the software is developed in a specific operating system and WINDOW facilities, the operating strategies can be applied globally.

• Journal of Korea Multimedia Society
• /
• v.8 no.12
• /
• pp.1581-1588
• /
• 2005

The primary goal of this paper is to facilitate the rate-distortion control in compressed domain, without introducing a full decoding and re-encoding system in pixel domain. For this aim, the error propagation behavior over several frame-sequences due to DCT coefficients-drop is investigated on the basis of statistical and empirical properties. Then, such properties are used to develop a simple estimation model for the CD distortion accounting for the characteristics of the underlying coded-frame. Experimental results show that the proposed model allows us to effectively control rate-distortions into coded-frames over different kinds of video sequences.

• The Proceeding of the Korean Institute of Electromagnetic Engineering and Science
• /
• v.6 no.2
• /
• pp.11-18
• /
• 1995

The goal of this research is to get a measuring system for absorption characteristics of electromagnetic wave absorbers in fernite type which are flat, grid-type and special type, etc. It is, however, very difficult to measure the absorption characteristics in as low frequency as 30 MHz. To solve the problem, therefore, we propose a standing wave method and a measurement method in time domain using paralledl striplines, construct the measuring system, and measure the charac- teristics of ferrite microwave absorbers in grid type using the proposed measuring system.