• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.05a
• /
• pp.367-369
• /
• 2022

In this paper, the predicted results were confirmed by label mapping for each class while implementing multi-class image classification to confirm accurate results for images in which the trained model failed classification. A CNN model was constructed and trained using Kaggle's Intel Image Classification dataset, and the mapped label values of multiple classes of images and the values classified by the model were compared by label mapping the images of the test dataset.

• Annual Conference on Human and Language Technology
• /
• 2020.10a
• /
• pp.333-336
• /
• 2020

개체명 인식이란 문장에서 인명, 지명, 기관명 등과 같이 고유한 의미를 갖는 단어를 찾아 개체명을 분류하는 작업이다. 딥러닝을 활용한 연구가 수행되면서 개체명 인식에 RNN(Recurrent Neural Network)과 CRF(Condition Random Fields)를 결합한 연구가 좋은 성능을 보이고 있다. 그러나 CRF는 시간 복잡도가 분류해야 하는 클래스(Class) 개수의 제곱에 비례하고, 최근 RNN과 Softmax 모델보다 낮은 성능을 보이는 연구도 있었다. 본 논문에서는 CRF의 단점을 보완한 LAN(Label Attention Network)와 사전 학습 언어 모델인 음절 단위 ELECTRA를 활용하는 개체명 인식 모델을 제안한다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.10a
• /
• pp.594-596
• /
• 2021

Future healthcare systems will heavily rely on ill-labeled data due to scarcity of the experts who are trained enough to label the data. Considering the contamination of the dataset, it is not desirable to make the neural network being overconfident to the dataset, but rather giving them some margins for the prediction is preferable. In this paper, we propose a novel epoch-wise decaying label-smoothing function to alleviate the model over-confidency, and it outperforms the neural network trained with conventional cross entropy by 6.0%.

• Proceedings of the Korean Institute of IIIuminating and Electrical Installation Engineers Conference
• /
• 1998.11a
• /
• pp.223-225
• /
• 1998

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.4
• /
• pp.992-999
• /
• 2014

This paper presents an extensive experimental comparison of a variety of multi-label learning methods for the accurate prediction of subcellular localization of proteins which simultaneously exist at multiple subcellular locations. We compared several methods from three categories of multi-label classification algorithms: algorithm adaptation, problem transformation, and meta learning. Experimental results are analyzed using 12 multi-label evaluation measures to assess the behavior of the methods from a variety of view-points. We also use a new summarization measure to find the best performing method. Experimental results show that the best performing methods are power-set method pruning a infrequently occurring subsets of labels and classifier chains modeling relevant labels with an additional feature. futhermore, ensembles of many classifiers of these methods enhance the performance further. The recommendation from this study is that the correlation of subcellular locations is an effective clue for classification, this is because the subcellular locations of proteins performing certain biological function are not independent but correlated.

• Journal of Life Science
• /
• v.8 no.6
• /
• pp.681-686
• /
• 1998

IASL(iodo acetamide) and MSL(maleimide) disordered the orderly helix arrangement of myosin in the rest state of spin level. Especially the effect of IASL was great. Equatorial reflection(10,11) change inferred that myosin head was moved to the vicinity of actin filament by spin level. The intensity change of 143 $\AA$ and 72 $\AA$ could offer infor-mation of the mass projection of population of myosin heads along the filament axis. The slope of intensity profile of the mass projection of 143 $\AA$ and reflection of IASL is appeared and that of MSL is appeared sharply. The dec-rease of 215 $\AA$ reflection intensity the periodical characteristic of 143 $\AA$ reflection by spin label. The raise of MSL actin reflection at 51 $\AA$ and 59 $\AA$ in the actin reflection change refers that the shifted myosin head binds a certain actin or changes an actin structure by spin label effect. Because iodo acetamide has a tendency to decease the actin reflection, actin dose not bind myosin head. From this result, we could conclude that LASL and MSL are spin labeled on SH of myosin head and disordered the helix arrangement of actin.

• Journal of the Korean Society of Food Culture
• /
• v.20 no.5
• /
• pp.525-531
• /
• 2005

This study was carried out to establish the consumer-centric food labeling system by investigating usage state, importance evaluation, problems and satisfaction, etc. on the current food labeling system by consumer. For usage state of checking the food label, 50.2% of respondents were replied ‘mostly check the label’, and they were indicated a significant difference on ‘education level(p<.05)’, and in case of married respondents, most were relied 'absolutely check the label', For the reason to check the food label, 61.8% of respondents were replied 'to determine whether it is stable or not', and they were indicated a significant difference on age(p<.01), marital status(p<.01), and job(p<.01). For the item considered as important things on the label of the whole food, 49.8% of respondents were replied that 'expiration and manufacture date is important', and they were indicated a significant difference(p<.001) on age, marital status, job. For item considered as important things on the label of each food, it was indicated that they considered food company as important thing in case of snack, soft drink, edible oils, and noodles, and food company in case of ice cream, and expiration and manufacture date in case of milk product and meat product. For the problems on the current food labeling system, the clauses 'Poor marking on food additives and materials contents' and 'untrue marking of nutrition contents' were indicated a significant difference on age(p<0.05, p<0.001), and the clause 'untrue marking of nutrition contents' were indicated a significant difference on marital status(p<.01) and job(p<.01).

• The Research Journal of the Costume Culture
• /
• v.23 no.3
• /
• pp.523-536
• /
• 2015

Corporate association-which refers to consumers' beliefs, knowledge, perceptions, and evaluations of a corporation -can affect consumers' purchasing intentions. Corporate association consists of corporate ability association and corporate social responsibility association. Corporate ability association refers to a company's product quality, corporate innovation, productivity, consumer orientation, and after service. Corporate social responsibility association, which refers to the social perspective a company has of its responsibility to society, can affect corporate image and consumers' purchasing intentions. Eco-labeling for protecting and sustaining the environment is one of the important green marketing strategies in the fashion business that can influence corporate association and consumers' purchasing intentions. The purpose of this study was to investigate the effect of consumers' eco-label recognition on their corporate association and intentions to purchase eco-friendly fashion products. Questionnaires were distributed to consumers. The 263 usable questionnaires that were returned were analyzed by descriptive statistics, Cronbach's alpha, factor analysis, regression analysis, and t-test. The results were as follows: There was a significant effect of eco-label recognition on corporate association (ability association and social responsibility association). Eco-label recognition and corporate association were found to significantly affect consumers' purchasing intentions. Regarding the eco-friendly fashion product buying experience, there was no significant difference on corporate association and buying intention, but there was significant difference on eco-label recognition.

• Journal of the Korean Dietetic Association
• /
• v.16 no.3
• /
• pp.226-238
• /
• 2010

This study was conducted to assess the effects of nutrition label education in children. The students of three randomly selected sixth-grade classes at an elementary school participated, were classified into a control group, an indirect (written message to parents) education group, and a direct (teaching children) education group. Nutrition label education was conducted in the latter two groups but not in the control group. There were no significant differences between pretest and posttest scores for nutrition knowledge or in the dietary attitudes of the control group and indirect education group. Improvements in test scores for nutrition knowledge and dietary attitudes were observed only in the direct education group. There were significant positive changes in the indirect and direct education groups, such as selecting snacks (cookies) by considering the fat content. Moreover, after the lessons, the percentage of children who chose white milk (regular milk, calcium-enriched milk, and low-fat milk) significantly increased in the direct education group. This study suggests that conducting nutrition label education helps children improve their nutrition knowledge and leads to positive changes in dietary attitudes and eating behaviors. In particular, a direct nutrition education method is more effective than an indirect method such as a written message to parents. Therefore, it is suggested that systematic and constant 'nutrition label education' be performed by nutrition teachers in elementary schools to establish healthy eating habits based on choosing nutritious snacks and processed foods.

• Journal of KIISE:Information Networking
• /
• v.30 no.6
• /
• pp.787-798
• /
• 2003

The flow label field in IPv6 has been proposed to provide the QoS. Since the existing flow label specification scheme like random-number format utilizes the label only as the identifier of flow, it is not appropriate for providing differentiated services according to the characteristics of various types of real-time traffic. This paper proposes a hybrid scheme that makes use of the flow label fields as components of flow and QoS parameters as well. To be specific, this paper investigates a scheme that both guarantees the end-to-end service quality and utilizes efficiently backbone resources by allowing users to specify QoS parameters using flow labels. Assuming an MLPS-TE network as the backbone, we compare the performance of our proposed scheme with that of random-number scheme through simulation. The simulation result shows that our scheme is more efficient than the existing one in terms of the transmission rate as well as the resource utilization of the backbone.