• The Journal of the Korea institute of electronic communication sciences
• /
• v.18 no.3
• /
• pp.421-426
• /
• 2023

Among artificial intelligence convergence technologies, Chatbot is an artificial intelligence-based interactive system and refers to a system that can provide interaction with humans. Chatbots are being re-examined as chatbots develop into NLP, NLU, and NLG. However, artificial intelligence chatbots can provide biased information based on learned data and cause serious damage such as privacy infringement and cybersecurity concerns, and it is essential to understand artificial intelligence technology and foster AI literacy. With the continued evolution and universalization of artificial intelligence, AI Literacy will also expand its scope and include new areas. This study is meaningful in raising awareness of artificial intelligence technology and proposing the use of human respect technology that is not buried in technology by cultivating human AI literacy capabilities.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.18 no.3
• /
• pp.449-458
• /
• 2023

Many people think that sign language is only for people who are deaf and cannot speak, but of course it is necessary for people who want to talk with them. One of the biggest challenges in ASL(American Sign Language) alphabet recognition is the high inter-class similarities and high intra-class variance. In this paper, we proposed an architecture that can overcome these two problems, which performs similarity learning to reduces inter-class similarities and intra-class variance between images. The proposed architecture consists of the same convolutional neural network with a double configuration that shares parameters (weights and biases) and also applies the Keras API to reduce similarity learning and variance through this pathway. The similarity learning results the use of the dual CNN shows that the accuracy is improved by reducing the similarity and variability between classes by not including the poor results of the two classes.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2022.05a
• /
• pp.97-97
• /
• 2022

AI 모형을 적용한 도시지역 침수예측에 대한 연구는 꾸준히 수행되어 왔다. AI 모형을 이용해 도시침수예측을 하기 위해서는 모형에 강우자료를 학습시키게 되는데, 시계열 강우분포 자료를AI 모형의 학습자료로 사용하기에 자료의 양이 너무 많기 때문에 총 강우량만을 이용하여 도시침수예측을 수행한 바 있다(Kim et al., 2021). 하지만 총 강우량만을 AI 모형에 학습시킬 경우, 지속기간 동안 강우가 고르게 분포하는지 불규칙적으로 분포하는지에 대한 정보가 포함되지 않았기 때문에 침수예측력이 떨어질 수 있다. 따라서 본 연구에서는 시계열 강우자료의 통계치를 산정하여 AI 모형에 학습시킴으로써 강우분포특성을 고려한 침수예측을 통해 예측력을 높이고자 한다. 총 강우량만을 학습시킬 경우, 같은 지속시간에 같은 양의 강우가 내리더라도 고른 분포를 가진 강우에 의해서는 실제 침수는 작게 일어나므로 과대예측을, 전체 지속시간 중 특정 시간대에 편향된 분포를 가진 강우에 의해서는 실제 침수가 크게 일어나므로 과소예측을 하는 문제가 발생할 수 있다. 따라서 표준편차를 평균 강우량으로 나눈 값인 변동계수, 강우분포의 뾰족한 정도를 나타내는 첨도, 평균값에 대해 어느 방향으로 비대칭인지를 나타내는 왜도 값을 추가로 학습시킴으로써 시계열 강우자료 전체를 학습시키지 않고도 강우분포를 학습시키지 않았을 때 발생하는 과소·과대예측 문제를 해결할 수 있다. 또한 변동계수 대신 표준편차를 학습시키는 모형, 변동계수와 표준편차를 모두 학습시키지 않는 모형, 변동계수와 표준편차를 모두 학습시키는 모형과의 침수예측 결과 비교를 통해 표준편차와 변동계수 중 어떤 통계치를 학습시키는 것이 적합한지와 비슷한 통계치 자료를 모두 학습시켰을 때의 과적합 문제 등에 대한 결론를 얻을 수 있다.

• Journal of Korea Society of Industrial Information Systems
• /
• v.29 no.2
• /
• pp.15-25
• /
• 2024

For effective analysis of animal ecosystems, technology that can automatically identify the current status of animal habitats is crucial. Specifically, animal sound classification, which identifies species based on their sounds, is gaining great attention where video-based discrimination is impractical. Traditional studies have relied on a single deep learning model to classify animal sounds. However, sounds collected in outdoor settings often include substantial background noise, complicating the task for a single model. In addition, data imbalance among species may lead to biased model training. To address these challenges, in this paper, we propose an animal sound classification scheme that combines predictions from multiple models using Focal Loss, which adjusts penalties based on class data volume. Experiments on public datasets have demonstrated that our scheme can improve recall by up to 22.6% compared to an average of single models.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.29 no.1
• /
• pp.81-89
• /
• 2011

The previous land gravity data in Korea showed locally biased irregular distribution. Especially, this problem was more serious in the mountainous area where the data density was significantly low. The same problem appeared in GPS/Levelling data thus the precision of the geoid could not be improved. From 2008, new gravity and GPS/Levelling data has been collected by the unified control point and survey on the benchmark project which were funded by the national geographic information institute. The newly obtained data has much better distribution and precision so that it could be used for update precision of geoid model. In this study, the new precision geoid has been calculated based old and new gravity data and this model showed 5.29cm of precision compared to 927 points of GPS/Levelling data. And the degree of fit and precision of hybrid geoid has been calculated 2.99cm and 3.67cm. The new gravimetric geoid has been updated about 27% over whole country. And it showed 42% of precision update due to collection of new gravity data on the Kangwon/Kyeongsang area which showed quite low distribution. In 2010, about 4,000 points of gravity and 300 points of GPS/Levelling data has been obtained by unified control and survey on benchmark project. We expect that new data will contribute to updating geoid precision and veri tying precision more objectively.

• The Journal of Korean society of community based occupational therapy
• /
• v.8 no.3
• /
• pp.77-89
• /
• 2018

Objective : The purpose of this study was to analyze occupational therapy intervention on the community mobility for stroke patients, and to provide evidence of intervention in the clinical fields. Methods : A systematic review was executed according to the PRISMA checklist. The accessed database was PubMed, EMBASE, Cochrane Library (CENTRAL), ProQuest Dissertations & thesis (PQDT), RISS, and KoreaMed. We included the articles published from 2005 to September 2018. RoBANS checklist was used to evaluate the quality of the articles. Included articles, totally eight, were categorized according to the type of intervention. Results : The study design of the literature was varied from two-group randomized trial, quasi-experimental study, case-control trial, one group pre-post comparison study, and cross-sectional study. In the evidence level, 6 articles were included in level II (75%). The percentage of low risk of bias in each article ranged from 52.5%~87.5%. Four studies (50%) provided intervention based on virtual reality or virtual environment. The three (37.5%) provided intervention based on the protocol, and the other (12.5%) did wheelchair training. All studies reported significant effects of the intervention. Conclusion : This systematic review provided evidences to use proper intervention in the clinical fields. Various type of studies should be conducted to prove the effect of occupational therapy intervention for community mobility.

• Journal of Nutrition and Health
• /
• v.54 no.5
• /
• pp.459-473
• /
• 2021

Purpose: Although numerous systematic reviews or meta-analysis have reported the hypotensive effects of garlic, the application of these results in the area of functional food is limited. This is because the trials used various garlic preparations and patients with differing hypertensive intensities. To validate the use of garlic powder as a blood pressure lowering functional food, we performed the current meta-analysis, focusing on the study of prehypertensive subjects. Methods: Literature search was carried out using various database up to July 2020, including PubMed, Cochrane, ScienceDirect and Korean studies Information Service System, and each study was screened by pre-stated inclusion/exclusion criteria. We identified nine trials that met the eligibility, of which two studies with moderate or high risk of bias were excluded. Results: Meta-analysis of the seven studies revealed that an intake of garlic powder significantly lowered the systolic blood pressure (SBP) and diastolic blood pressure (DBP) by -6.0 mmHg (95% confidence interval [CI], -11.2, -0.8; p = 0.025) and -2.7 mmHg (95% CI, -5.3, -0.1; p = 0.046), respectively. Shapes of the funnel plot for both SBP and DBP seemed symmetrical, and the Egger's regression revealed no publication bias. Moreover, duration of the intervention period was inversely associated with the pooled effects of garlic powder on SBP (p = 0.019) and DBP (p = 0.019), and this result was supported by the subgroup-analysis. The daily dose of garlic powder, baseline value of each biomarker, and subject number, did not moderate the effects on SBP and DBP. Conclusion: Results of the present meta-analysis indicate that garlic powder supplements are superior to placebo for improving the BP in prehypertensive individuals.

• Journal of Intelligence and Information Systems
• /
• v.29 no.3
• /
• pp.317-340
• /
• 2023

In a business context, ESG is defined as the use of environmental, social, and governance factors to assess a firm's progress in terms of sustainability. Social media has enabled the public to actively share firms' good and/or bad deeds, increasing public interest in ESG management. Therefore, this study aimed to investigate the association of firm performances with the respective sentiments towards each of environmental, social, and governance activities, as well as comprehensive ESG sentiments, which encompass all environmental, social, and governance sentiments. This study used panel regression models to examine the relationship between social media ESG sentiment and the Return on Assets (ROA) and Return on Equity (ROE) of 143 companies listed on the KOSPI 200. We collected data from 2018 to 2021, including sentiment data from a variety of social media channels, such as online communities, Instagram, blogs, Twitter, and other news. The results indicated that firm performance is significantly related to respective ESG and comprehensive ESG sentiments. This study has several implications. By using data from various social media channels, it presents an unbiased view of public ESG sentiment, rather than relying on ESG ratings, which may be influenced by rating agencies. Furthermore, the findings can be used to help firms determine the direction of their ESG management. Therefore, this study provides theoretical and practical insights for researchers and firms interested in ESG management.

• The Korean Journal of Applied Statistics
• /
• v.29 no.7
• /
• pp.1459-1473
• /
• 2016

Both longitudinal data and survival data are collected simultaneously in longitudinal data which are observed throughout the passage of time. In this case, the effect of the independent variable becomes biased (provided that sole use of longitudinal data analysis does not consider the relation between both data used) if the missing that occurred in the longitudinal data is non-ignorable because it is caused by a correlation with the survival data. A joint model of longitudinal data and survival data was studied as a solution for such problem in order to obtain an unbiased result by considering the survival model for the cause of missing. In this paper, a joint model of the longitudinal zero-inflated count data and survival data is studied by replacing the longitudinal part with zero-inflated count data. A hurdle model and proportional hazards model were used for each longitudinal zero inflated count data and survival data; in addition, both sub-models were linked based on the assumption that the random effect of sub-models follow the multivariate normal distribution. We used the EM algorithm for the maximum likelihood estimator of parameters and estimated standard errors of parameters were calculated using the profile likelihood method. In simulation, we observed a better performance of the joint model in bias and coverage probability compared to the separate model.

• Survey Research
• /
• v.10 no.3
• /
• pp.157-172
• /
• 2009

For a survey research in which the characteristics of the population of interest are investigated from a sample, representativeness of the sampling frame is one of the most important part to be considered. If the sampling frame fails to represent the population properly, statistical procedures based on the even efficient sampling design result in significant nonsampling biases and thus the statistical validities of the results could be damaged. But the construction of the reliable sampling frame that covers the population properly costs money and time and thus the sampling frame based on a census or a large scale survey is often used in practice. For example, the sampling frame based on the population households census is used for many household surveys in Korea. But due to the time difference between the census and a survey of interest, the sampling frame constructed from the census is expected to fail to cover the population of interest. Especially, one could expect a large amount of population and household movement in a large city like Seoul. Thus in our research, we considered the construction of new sampling frame and the procedure of sample selection for 2008 Seoul survey. We analyzed the sampling frame based on 2005 population households census and found that it does not represent the population properly. Thus, we proposed a new sampling frame based on resident registration DB for 2008 Seoul survey. We also proposed the sampling weights and estimator of the population mean based on the sample selected from the newly constructed sampling frame.