• The Journal of the Korea institute of electronic communication sciences
• /
• v.16 no.3
• /
• pp.401-406
• /
• 2021

Recently, mean-teacher models based on convolutional recurrent neural networks are popularly used in audio event detection. The mean-teacher model is an architecture that consists of two parallel CRNNs and it is possible to train them effectively on the weakly-labelled and unlabeled audio data by using the consistency learning metric at the output of the two neural networks. In this study, we tried to improve the performance of the mean-teacher model by using additional derivative features of the log-mel spectrum. In the audio event detection experiments using the training and test data from the Task 4 of the DCASE 2018/2019 Challenges, we could obtain maximally a 8.1% relative decrease in the ER(Error Rate) in the mean-teacher model using proposed derivative features.

• Korean Journal of Childcare and Education
• /
• v.18 no.2
• /
• pp.39-56
• /
• 2022

Objective: The purpose of this study was to investigate the influence of perceived supervisor's servant leadership and teacher self-efficacy for positive teacher-child relationships on work engagement using the path model. Methods: The subjects were 210 teachers working at childcare centers in Daegu and Gyeongbuk Province. Questionnaires were used to investigate perceived supervisor's servant leadership, teacher self-efficacy for positive teacher-child relationships, and work engagement. The collected data were analyzed by Structural Equation Modeling(SEM), Bootstrapping, Pearson Correlation, AMOS 20.0, and SPSS 21.0. Results: First, supervisor's servant leadership had a positive influence on teacher self-efficacy for positive teacher-child relationships. Second, teacher self-efficacy for positive teacher-child relationships had a positive influence on work engagement. Third, supervisor's servant leadership had a positive influence on teacher's work engagement. Fourth, supervisor's servant leadership had an indirec effect on teacher's work engagement by teacher self-efficacy for positive teacher-child relationships. Conclusion/Implications: The results mean teacher self-efficacy for positive teacher-child relationships had more influence on teacher's work engagement. Therefore, it is more useful to promote teacher self-efficacy for positive teacher-child relationships in order to strengthen teacher's work engagement.

• The Journal of the Acoustical Society of Korea
• /
• v.42 no.5
• /
• pp.395-401
• /
• 2023

Sound Event Detection (SED) aims to identify not only sound category but also time interval for target sounds in an audio waveform. It is a critical technique in field of acoustic surveillance system and monitoring system. Recently, various models have introduced through Detection and Classification of Acoustic Scenes and Events (DCASE) Task 4. This paper explored how to design optimal parameters of DenseNet based model, which has led to outstanding performance in other recognition system. In experiment, DenseRNN as an SED model consists of DensNet-BC and bi-directional Gated Recurrent Units (GRU). This model is trained with Mean teacher model. With an event-based f-score, evaluation is performed depending on parameters, related to model architecture as well as model training, under the assessment protocol of DCASE task4. Experimental result shows that the performance goes up and has been saturated to near the best. Also, DenseRNN would be trained more effectively without dropout technique.

• Korean Journal of Child Studies
• /
• v.37 no.4
• /
• pp.145-158
• /
• 2016

Objective: This study investigated the moderating effect of emotional dysregulation on the relationship between teacher efficacy and job stress among teachers in the Early Childhood Education and Care (ECEC). Methods: The participants included 586 ECEC teachers from 99 centers in Seoul, Gyeonggi-do, Daejeon, Chungcheong-do, Jeolla-do, and Gyeongsang-do. The data were analyzed through frequencies, percentages, and Pearson's correlations using SPSS 21.0 (IBM Co., Armonk, NY). To analyze the moderating effect, Ping's (1996) two-step approach was used via AMOS 20.0 (IBM Co., Armonk, NY). Results: The main findings are as follows. First, the mean scores of ECEC teacher efficacy and job stress showed above the average, and the mean score of emotional dysregulation was the nearly average. Second, fit statistics indicated that the proposed model, as revised, provided an acceptable fit to the sample data. This proposed model showed that the emotional dysregulation of teachers in ECEC had a moderating effect on the relationship between teacher efficacy and job stress. Conclusion: These findings imply that the ECEC teachers showed the higher level of self-trust and self-confidence than average regarding their own work, and suffered from the work overload. Also, the positive and supportive working environment would help the ECEC teachers to reduce their emotional dysregulation. In addition, there was a moderating effect of the ECEC teachers' emotional dysregulation on the relationship between teacher efficacy and job stress. These findings imply that the workshop or counselling programs need to be provided to teachers in order to help control their emotion dysregulation and reduce their job stress.

• Journal of the Korean Society for Library and Information Science
• /
• v.17
• /
• pp.229-259
• /
• 1989

The school library plays very important and central role for the fullfillment of educational objective in the process of implementing the school curriculum. And the media specialist known as teacher-librarian is. primarily responsible for the successful operation and effective management of the media center. It is true that effectiveness of the school library depends greatly on the cooperative coordination among teacher-librarian, teacher and principal in addition to the strong support and effective supervision of school district authority, specifically, supervisor of school library program. The successful program therefore is influenced by their perception of the teacher-librarian's role. The purpose of this study was to investigate the teacher-librarian's. educational role perceived by teacher-librarian, teacher, principal and school district supervisor. A standard task model describing both the management and educational functions and expectations of teacher-librarian was proposed after studying the relevant current standards of four countries. Serveral hypotheses were made regarding role perception and a survey instrument was designed based on this proposed task model and the results of the literature review to validate the hypotheses. The questionnaires were mailed out and the returned questionaires were analyzed by SPSS computer routine. The major findings are summarized as follows: 1) There was difference of the role perception among teacher-librarian, teacher, principal and district supervisor at the significance level of p<0.001 for both management and educational functions of teacher­librarian. The mean score of the teacher-librarian was highest among four groups and the score for the educational function tasks were lower than that of management function tasks. 2) The positive correlation exists at the significance level of p<0.001 between the role perception and the attitude toward changing social trends such as acceptance of new technology, need for better communication and information. 3) The positive correlation exists at the significance level of p<0.001 between the role perception and the view on the curriculum and teaching method.

• Journal of Korean Home Economics Education Association
• /
• v.13 no.3
• /
• pp.1-11
• /
• 2001

The purpose in this study was to investigate stages of concern of home economics teachers about the practical reasoning instruction(PRI) innovation configuration. levels of use. using Concerns Based Adoption Model(CBAM) The design of the study was descriptive. The Questionnaires of stages of concern. levels of use. and innovation configuration developed by Hall in 1987 were used. Data from HE teachers were collected through mailing. focus interview. and phone calls. Mean. percents. an frequencies were used to describe stages of concern, level of use. and innovation configuration. The results of the study were as follows: 1. The highest number of HE teacher respondents was stage 1. Information in March and july. 2 In terms of mean of PRI configuration. percentile. the HE teachers had implemented over 63% of the PRI elements and the cooperative learning elements in both March and July. However. of evaluation elements, 33% in March and 47% in July had been conducted. 3. The highest number of HE teacher respondents was level 3. Mechanical level in March. and July.

• Korean Journal of Childcare and Education
• /
• v.15 no.2
• /
• pp.1-19
• /
• 2019

Objective: This study aims to examine the factors associated with teacher intention to report child abuse in child care centers. Methods: Data were collected from teachers at child care centers in the Jeonbuk region, using a self-administered questionnaire. In this study, 186 completed questionnaires were used to analyze the data. Results: The respondents reported their intention to report child abuse as follows: physical abuse(49.5%), emotional abuse(16.1%), sexual abuse(12.9%), and neglect(15.6%). The multiple hierarchical regression analyses revealed that participation at child abuse prevention training programs and awareness of reporting procedures were significantly associated with teacher intention to report child abuse at child care centers in Model 2. Also, attitudes towards reporting child abuse were significantly associated with teacher intention to report child abuse in Model 3. Conclusion/Implications: This study suggests evaluating chid abuse prevention training programs more accurately because the respondents who did not participate in the programs showed statistically significant higher mean scores of intention to report child abuse than who participated(1.83 vs .85). In addition, educational programs about child abuse for teachers in child care centers need to focus on changes in attitudes towards reporting child abuse, which in turn can change behavior.

• Journal of the Korean Data and Information Science Society
• /
• v.20 no.5
• /
• pp.905-912
• /
• 2009

We propose a doubly penalized kernel machine (DPKM) which uses heteroscedastic location-scale model as basic model and estimates both mean and volatility functions simultaneously by kernel machines. We also present the model selection method which employs the generalized approximate cross validation techniques for choosing the hyperparameters which affect the performance of DPKM. Artificial examples are provided to indicate the usefulness of DPKM for the mean and volatility functions estimation.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.18 no.2
• /
• pp.51-58
• /
• 2023

This paper proposes a deep learning method for estimating the heart rate from facial videos. Our proposed method estimates remote photoplethysmography (rPPG) signals to predict the heart rate. Although there have been proposed several methods for estimating rPPG signals, most previous methods can not be utilized in low-power single board computers due to their computational complexity. To address this problem, we construct a lightweight student model and employ a knowledge distillation technique to reduce the performance degradation of a deeper network model. The teacher model consists of 795k parameters, whereas the student model only contains 24k parameters, and therefore, the inference time was reduced with the factor of 10. By distilling the knowledge of the intermediate feature maps of the teacher model, we improved the accuracy of the student model for estimating the heart rate. Experiments were conducted on the UBFC-rPPG dataset to demonstrate the effectiveness of the proposed method. Moreover, we collected our own dataset to verify the accuracy and processing time of the proposed method on a real-world dataset. Experimental results on a NVIDIA Jetson Nano board demonstrate that our proposed method can infer the heart rate in real time with the mean absolute error of 2.5183 bpm.

• Journal of the Korea institute for structural maintenance and inspection
• /
• v.27 no.5
• /
• pp.113-119
• /
• 2023

Most infrastructure structures were completed during periods of economic growth. The number of infrastructure structures reaching their lifespan is increasing, and the proportion of old structures is gradually increasing. The functions and performance of these structures at the time of design may deteriorate and may even lead to safety accidents. To prevent this repercussion, accurate inspection and appropriate repair are requisite. To this end, demand is increasing for computer vision and deep learning technology to accurately detect even minute cracks. However, deep learning algorithms require a large number of training data. In particular, label images indicating the location of cracks in the image are required. To secure a large number of those label images, a lot of labor and time are consumed. To reduce these costs as well as increase detection accuracy, this study proposed a learning structure based on mean teacher method. This learning structure was trained on a dataset of 900 labeled image dataset and 3000 unlabeled image dataset. The crack detection network model was evaluated on over 300 labeled image dataset, and the detection accuracy recorded a mean intersection over union of 89.23% and an F1 score of 89.12%. Through this experiment, it was confirmed that detection performance was improved compared to supervised learning. It is expected that this proposed method will be used in the future to reduce the cost required to secure label images.