• The Journal of the Acoustical Society of Korea
• /
• v.41 no.3
• /
• pp.335-341
• /
• 2022

In this paper, we adopt a transformer that shows remarkable performance in natural language processing as an acoustic model of hybrid speech recognition. The transformer acoustic model uses attention structures to process sequential data and shows high performance with low computational cost. This paper proposes a method to improve the performance of transformer AM by applying each of the four algorithms of sequence discriminative training, a weighted finite-state transducer (wFST)-based learning used in the existing DNN-HMM model. In addition, compared to the Cross Entropy (CE) learning method, sequence discriminative method shows 5 % of the relative Word Error Rate (WER).

• Child Health Nursing Research
• /
• v.30 no.2
• /
• pp.97-107
• /
• 2024

Purpose: This study aimed to investigate the impact of family resilience on caregiver well-being across various age groups of children diagnosed with ADHD (attention-deficit/hyperactivity disorder) groups (0-5, 6-11, and 12-17 years). Methods: Utilizing secondary data from the 2022 US National Survey of Child Health, this cross-sectional study involved 2,752 children who were formally diagnosed with ADHD. Statistical analysis included descriptive analysis, Spearman's rank correlation, chi-square tests, and linear regression, conducted using SPSS version 27. Results: The study revealed a moderate positive correlation (r=.35, p<.001) between family resilience and caregiver well-being. Controlling for covariates, family resilience accounted for 25.2%, 21.1%, and 22.1% of caregiver well-being variance in age groups 0-5, 6-11, and 12-17 years, respectively. Additionally, factors like employment status, family structure, and caregiver age showed varying influences on caregiver well-being across these developmental stages. A consistent pattern emerged across these age groups: unemployment and non-traditional family structures were associated with negative impacts on caregiver well-being, whereas older caregiver age positively influenced well-being. Conclusion: This research underscores the importance of age-specific family resilience strategies to improve caregiver well-being and family interactions in ADHD contexts. Investigating these aspects through qualitative studies across various cultures could deepen our understanding of well-being and inform culturally sensitive interventions.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.9
• /
• pp.4665-4683
• /
• 2019

Previous methods build image annotation model by leveraging three basic dependencies: relations between image and label (image/label), between images (image/image) and between labels (label/label). Even though plenty of researches show that multiple dependencies can work jointly to improve annotation performance, different dependencies actually do not "work jointly" in their diagram, whose performance is largely depending on the result predicted by image/label section. To address this problem, we propose the adaptive attention annotation model (AAAM) to associate these dependencies with the prediction path, which is composed of a series of labels (tags) in the order they are detected. In particular, we optimize the prediction path by detecting the relevant labels from the easy-to-detect to the hard-to-detect, which are found using Binary Cross-Entropy (BCE) and Triplet Margin (TM) losses, respectively. Besides, in order to capture the inforamtion of each label, instead of explicitly extracting regional featutres, we propose the self-attention machanism to implicitly enhance the relevant region and restrain those irrelevant. To validate the effective of the model, we conduct experiments on three well-known public datasets, COCO 2014, IAPR TC-12 and NUSWIDE, and achieve better performance than the state-of-the-art methods.

• Journal of Korean Public Health Nursing
• /
• v.25 no.2
• /
• pp.221-232
• /
• 2011

Purpose: The study sought to provide basic information for North Korean defectors (Sae-Tu-Min) by studying the quality of life(QoL) based on their general characteristics and health problems. Methods: The targeted subjects were 1400 North Korean settlers who were receiving settlement service by Korean National Red Cross after finishing their training at Hanawon in 2005. The data was collected from August to October in 2007. The QoL was assessed by Medical Outcomes Study Short Form-36. Collected data were analyzed by ANOVA and t-test with SPSS win 15.0 program. Result: The majority of the 466 subjects were female(77.9%). The average age was 36.8 years, 52.8% lived alone, and 64.0% were unemployed. The average QoL score of the group was $62.4{\pm}18.4$, which was markedly lower in terms of social, physical and mental health compared to the scores of immigrants to South Korea from other countries. The score was the highest for physical ability and the lowest for general health. Influential socio-demographic factors to QoL were gender, age, education, religion, annual salary, occupation, and life insurance. Employed in the subject group showed higher level of QoL on eight categories than unemployed. Concerning health problems that detracted from QoL, 18.5% of the subjects were aware they suffered from anemia and 8.8% of them recognized they suffered from attention deficits. Conclusion: The health conditions of the North Korean settlers had improved since they settled in South Korea, but their QoL was lower than other immigrants. Various programs to maintain and promote the health of North Korean settlers would be useful.

• Journal of the Korean Society of Food Culture
• /
• v.29 no.5
• /
• pp.464-475
• /
• 2014

Eating breakfast provides crucial nutrition for brain function and helps promote overall health. It is especially critical in growing adolescents, as it is known to form good eating habits and better study habits. This study investigated the effects of skipping breakfast on nutritional state, fatigue level, and attention level. A cross-sectional study was conducted in 2010 on total of 828 adolescents composed of 414 boys and 414 girls. Students who ate breakfast never to twice per week were placed in the breakfast-skipper group while students who ate breakfast more than five times per week were included in the breakfast-eater group. Students performed a self-reported questionnaire on food behaviors, amount of food consumption, fatigue level, attention deficient hyperactivity disease (ADHD) level by Conners-Wells' Adolescent Self-Report Scales, depression scale, and self-esteem level. Statistical analysis was conducted using the SAS program (version 9.1). A total of 135 boys (32.6%) and 138 girls (33.3%) were included in the breakfast-skipper group, whereas 241 boys (58.2%) and 223 girls (53.9%) were included in the breakfast-eater group. The breakfast-skipper group showed irregular food behaviors and lacked nutrients. Specifically, energy (p< .001), protein (p< .001), dietary fiber (p< .001), calcium (p< .01), vitamin A (p< .01), thiamin (p< .05), niacin (p< .001) levels in boy breakfast-skippers were statistically lower compared to boy breakfast-eaters. Intakes of all nutrients except fat in girl breakfast-skippers were statistically lower than in girl breakfast-eaters. Girl breakfast-skippers (41.3%) showed significantly higher fatigue risks compared to girl breakfast-eaters (21.5%). Low attention level was also observed only in girls in the breakfast-skipping group. Moreover, students in the breakfast-skipper group showed higher scores for depression and low self-esteem (p< .001). In conclusion, skipping breakfast has effects on young adolescents' nutrition, manifesting as high fatigue level and low attention level, especially in girls.

• The KSFM Journal of Fluid Machinery
• /
• v.17 no.5
• /
• pp.19-26
• /
• 2014

The cross flow turbine attracts more and more attention for its relatively wide operating range and simple structure. In this study, a novel type of micro cross flow turbine is developed for application to a step in an irrigational channel. The head of the turbine is only H=4.3m and the turbine inlet channel is open ducted type, which has barely been studied. The efficiency of the turbine with inlet open duct channel is relatively low. Therefore, a guide nozzle on the turbine inlet is attached to improve the performance of the turbine. The guide nozzle shapes are investigated to find the best shape for the turbine. The guide nozzle plays an important role on directing flow at the runner entry, and it also decreases the negative torque loss by reducing the pressure difference in Region 1. There is 12.5% of efficiency improvement by attaching a well shaped guide nozzle on the turbine inlet.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.7 no.12
• /
• pp.3071-3095
• /
• 2013

Configuring optimization of wireless sensor networks, which can improve the network performance such as utilization efficiency and network lifetime with minimal energy, has received considerable attention in recent years. In this paper, a cross layer optimal approach is proposed for multi-source linear network and grid network under Rayleigh block-fading channels, which not only achieves an optimal utility but also guarantees the end-to-end reliability. Specifically, in this paper, we first strictly present the optimization method for optimal nodal number $N^*$, nodal placement $d^*$ and nodal transmission structure $p^*$ under constraints of minimum total energy consumption and minimum unit data transmitting energy consumption. Then, based on the facts that nodal energy consumption is higher for those nodes near the sink and those nodes far from the sink may have remaining energy, a cross layer optimal design is proposed to achieve balanced network energy consumption. The design adopts lower reliability requirement and shorter transmission distance for nodes near the sink, and adopts higher reliability requirement and farther transmission distance for nodes far from the sink, the solvability conditions is given as well. In the end, both the theoretical analysis and experimental results for performance evaluation show that the optimal design indeed can improve the network lifetime by 20-50%, network utility by 20% and guarantee desire level of reliability.

• Journal of Computational Design and Engineering
• /
• v.2 no.2
• /
• pp.88-95
• /
• 2015

Recently, fiber composite materials have been attracting attention from industry because of their remarkable material characteristics, including light weight and high stiffness. However, the costs of products composed of fiber materials remain high because of the lack of effective manufacturing and designing technologies. To improve the relevant design technology, this paper proposes a novel simulation method for deforming fiber materials. Specifically, given a 3D model with constant thickness and known fiber orientation, the proposed method simulates the deformation of a model made of thick fiber-material. The method separates a 3D sheet model into two surfaces and then flattens these surfaces into two dimensional planes by a parameterization method with involves cross vector fields. The cross vector fields are generated by propagating the given fiber orientations specified at several important points on the 3D model. Integration of the cross vector fields gives parameterization with low-stretch and low-distortion.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.42 no.5 s.305
• /
• pp.123-136
• /
• 2005

We propose a recognition system for superimposed patterns based on selective attention model and SVM which produces better performance than artificial neural network. The proposed selective attention model includes attention layer prior to SVM which affects SVM's input parameters. It also behaves as selective filter. The philosophy behind selective attention model is to find the stopping criteria to stop training and also defines the confidence measure of the selective attention's outcome. Support vector represents the other surrounding sample vectors. The support vector closest to the initial input vector in consideration is chosen. Minimal euclidean distance between the modified input vector based on selective attention and the chosen support vector defines the stopping criteria. It is difficult to define the confidence measure of selective attention if we apply common selective attention model, A new way of doffing the confidence measure can be set under the constraint that each modified input pixel does not cross over the boundary of original input pixel, thus the range of applicable information get increased. This method uses the following information; the Euclidean distance between an input pattern and modified pattern, the output of SVM, the support vector output of hidden neuron that is the closest to the initial input pattern. For the recognition experiment, 45 different combinations of USPS digit data are used. Better recognition performance is seen when selective attention is applied along with SVM than SVM only. Also, the proposed selective attention shows better performance than common selective attention.

• Journal of Ocean Engineering and Technology
• /
• v.29 no.4
• /
• pp.317-321
• /
• 2015

Many kinds of generation systems have been developed to use ocean energy. Among these, with the use of an oscillating water column (OWC) for power generation is attracting attention. The OWC-type wave power generation system converts wave energy into electricity by operating a generator turbine with the oscillating water level in a column of water. There are two ways to convert wave power into electricity using an OWC. One uses a cross-flow turbine using the water level inside the OWC. The other method uses the flow of air in a Wells turbine, which depends on the water level. An experiment was carried out using a 2-D wave tank in order to minimize the number of empirical tests. The design factors were taken from Koo et al. (2012) and the experimental environment assumed by free surface motion. This paper deals with characteristics of two types of wave energy conversion systems combine with a breakwater. One model uses an air-driven Wells turbine and a cross-flow water turbine. The other type uses a cross-flow water turbine. Wave energy converters with OWCs have mostly been studied using air-driven Wells turbines. The efficiency of the cross-flow turbine was about 15% higher than that of the other model, and the water level of the OWC internal chamber for the cross-flow water turbine and air-driven Wells turbine was less than about 40% lower than the one using only the cross-flow water turbine.