• Journal of Korea Society of Industrial Information Systems
• /
• v.27 no.4
• /
• pp.1-10
• /
• 2022

Attempts to interpret human perspectives using computer vision have been developed in various fields. In this paper, we propose a method for evaluating the walking environment through semantic segmentation results of images from road images. First, the Kakao Map API was used to collect road images, and four-way images were collected from about 50,000 points in JeonJu. 20% of the collected images build datasets through crowdsourcing-based paired comparisons, and train various regression models using paired comparison data. In order to derive the walkability score of the image data, the ranking score is calculated using the Trueskill algorithm, which is a ranking algorithm, and the walkability and analysis using various regression models are performed using the constructed data. Through this study, it is shown that the walkability of Jeonju can be evaluated and scores can be derived through the correlation between pixel distribution classification information rather than human vision.

• The Journal of the Korea Contents Association
• /
• v.21 no.7
• /
• pp.510-518
• /
• 2021

This study analyzed (1) the utility of telepractice in language assessment and (2) parental satisfaction for telepractice. Young children with language disabilities conducted REVT through both face-to-face and telepractice system. The correlation and paired t-test were analyzed. The results showed (1) the strong correlations between telepractice and face-to-face assessment scores. However, the telepractice score was significantly lower than face-to-face in the receptive vocabulary test and (2) the overall average of parental satisfaction for telepractice was 3.6 out of 5 points. The parents showed the highest satisfaction in terms of time and convenience of telepractice while the equivalence between telepractice and face-to-face item was the lowest. This study suggested the possibility and considerations in applying telepractice of language assessment in Korea.

• The Korean Journal of Applied Statistics
• /
• v.36 no.4
• /
• pp.323-335
• /
• 2023

In causal analysis of high dimensional data, it is important to reduce the dimension of covariates and transform them appropriately to control confounders that affect treatment and potential outcomes. The augmented inverse probability weighting (AIPW) method is mainly used for estimation of average treatment effect (ATE). AIPW estimator can be obtained by using estimated propensity score and outcome model. ATE estimator can be inconsistent or have large asymptotic variance when using estimated propensity score and outcome model obtained by parametric methods that includes all covariates, especially for high dimensional data. For this reason, an ATE estimation using an appropriate dimension reduction method and semiparametric model for high dimensional data is attracting attention. Semiparametric method or sparse sufficient dimensionality reduction method can be uesd for dimension reduction for the estimation of propensity score and outcome model. Recently, another method has been proposed that does not use propensity score and outcome regression. After reducing dimension of covariates, ATE estimation can be performed using matching. Among the studies on ATE estimation methods for high dimensional data, four recently proposed studies will be introduced, and how to interpret the estimated ATE will be discussed.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.07a
• /
• pp.541-544
• /
• 2023

LLM(Largescale Language Model)의 성능 향상을 위한 비용 효율적인 방법으로 ChatGPT, GPT-4와 같은 초거대 모델의 output에 대해 SLM(Small Language Model)을 finetune하는 방법이 주목받고 있다. 그러나, 이러한 접근법은 주로 범용적인 지시사항 모델을 위한 학습 방법으로 사용되며, 제한된 특정 도메인에서는 추가적인 성능 개선의 여지가 있다. 본 연구는 특정 도메인(Writing Assistant)에서의 성능 향상을 위한 새로운 방법인 Self-Guided Approach를 제안한다. Self-Guided Approach는 (1) LLM을 활용해 시드 데이터에 대해 도메인 특화된 metric(유용성, 관련성, 정확성, 세부사항의 수준별) 점수를 매기고, (2) 점수가 매겨진 데이터와 점수가 매겨지지 않은 데이터를 모두 활용하여 supervised 방식으로 SLM을 미세 조정한다. Vicuna에서 제안된 평가 방법인, GPT-4를 활용한 자동평가 프레임워크를 사용하여 Self-Guided Approach로 학습된 SLM의 성능을 평가하였다. 평가 결과 Self-Guided Approach가 Self-instruct, alpaca와 같이, 생성된 instruction 데이터에 튜닝하는 기존의 훈련 방법에 비해 성능이 향상됨을 확인했다. 다양한 스케일의 한국어 오픈 소스 LLM(Polyglot1.3B, PolyGlot3.8B, PolyGlot5.8B)에 대해서 Self-Guided Approach를 활용한 성능 개선을 확인했다. 평가는 GPT-4를 활용한 자동 평가를 진행했으며, Korean Novel Generation 도메인의 경우, 테스트 셋에서 4.547점에서 6.286점의 성능 향상이 발생했으며, Korean scenario Genration 도메인의 경우, 테스트 셋에서 4.038점에서 5.795 점의 성능 향상이 발생했으며, 다른 유사 도메인들에서도 비슷한 점수 향상을 확인했다. Self-Guided Approach의 활용을 통해 특정 도메인(Writing Assistant)에서의 SLM의 성능 개선 가능성을 확인했으며 이는 LLM에 비용부담을 크게 줄이면서도 제한된 도메인에서 성능을 유지하며, LLM을 활용한 응용 서비스에 있어 실질적인 도움을 제공할 수 있을 것으로 기대된다.

• Journal of Practical Engineering Education
• /
• v.15 no.1
• /
• pp.193-198
• /
• 2023

This paper analyzed the validity of exercise problems for each unit in Python programming education. Practice questions presented for each unit are presented through an online learning system, and each student uploads an answer code and is automatically graded. Data such as students' mid-term exam scores, final exam scores, and practice questions scores for each unit are collected through Python lecture that lasts for one semester. Through the collected data, it is possible to improve the exercise problems for each unit by analyzing the validity of the automatic scoring exercise problems. In this paper, Orange machine learning tool was used to analyze the validity of automatic scoring exercises. The data collected in the Python subject are analyzed and compared comprehensively by total, top, and bottom groups. From the prediction accuracy of the machine learning model that predicts the student's final grade from the Python unit-by-unit practice problem scores, the validity of the automatic scoring exercises for each unit was analyzed.

• Journal of the Korean Applied Science and Technology
• /
• v.40 no.5
• /
• pp.977-987
• /
• 2023

The purpose of this study was to investigate the effect of visual information on the dynamic balance on Y-balance Test(YBT). The subjects of the study were 18 male and female adults in their 20s and 30s (age: 23.17±1.72 years, height: 172.46±9.84 cm, weight: 73.39±11.44 kg, leg length: 88.89±5.69 cm) who participated in the study. To measure dynamic balance between binocular and monocular use, absolute reach distance, composite score, and COP variables were measured on left and right feet through YBT and results were derived. As a result of the study, monocular block(left and right eye block) showed higher absolute reach and composite scores than binocular use in posterolateral, posteromedial, and composite scores during YBT. As a result of COP, there was no difference in anterior and posteromedial reach. When reaching posterolateral, AP COP velocity of left foot in monocular block appeared slower than that in binocular vision, and in COP velocity, COP velocity of left foot in monocular block appeared slower than binocular vision.

• Journal of The Korean Association For Science Education
• /
• v.28 no.6
• /
• pp.675-684
• /
• 2008

This study investigated the effects of the prescribed instructional strategy for reducing students' connecting errors in learning chemistry concepts with multiple external representations by students' field independence-dependence. Seventh graders (N=126) at a coed middle school were assigned to control and treatment groups. The students learned "Boyle's Law" and "Charles's Law" for two class periods. Results revealed that the students in the treatment group scored significantly higher than those in the control group in a conception test. The scores of the treatment group were significantly higher than those of the control group in a motivational learning test, especially in 'attention' of the test. However, there was no significant interaction between the instruction and students' field independence-dependence in the two tests. Most students in the treatment group perceived the instruction positively in cognitive and motivational aspects.

• Journal of the Korean Applied Science and Technology
• /
• v.40 no.6
• /
• pp.1249-1258
• /
• 2023

The purpose of this study was to reveal the effect of COP variable of support feet on relative reach distance and composite score of active feet during YBT. 29 adults in their 20s(age: 24.4±3.0 yrs, height: 171.0±10.5 cm, weight: 72.1±12.1 kg, leg length: 88.2±5.8 cm) participated in the study. Using YBT, relative reach distance and composite score of active feet were calculated, and COP variables of support feet were measured on left and right support feet. Multiple regression analysis was used to determine the effect of COP variables of support feet during YBT on relative reach distance and composite score. As a result of the study, during YBT, when supporting right feet, the influence of AP COP velocity, ML COP velocity and COP velocity in posterolateral direction was significant, and when supporting left feet, the influence of AP COP velocity, ML COP velocity in posteromedial direction improved relative reach distance and composite score.

• Journal of dental hygiene science
• /
• v.11 no.3
• /
• pp.265-272
• /
• 2011

The purpose of this study is to compare oral hygiene status between young adult group and older adult group and to suggest a pathway model of oral health level in self-assessed. Subjects were totally 80 people with young adult group and the older adult group. The data, which was collected by surveying for one month from July 1, 2010 to July 30, was analyzed with SPSS VERSION 19.0 and AMOS 19.0. Oral hygiene test and questionnaire survey were carried out from targeting. 1. The young adult group and older adult group had the same quality in oral health level score and living habit. 2. The oral hygiene status in young adult and older adult group was same in 7 items(58.33%) out of 12 items. 3. Oral health level score in self-assessed of young adult group could be explained with 34% by 3 variables. Older adult group could be explained with 26% by 2 variables. 4. In young adult, the fitness of path model was very good with 0.96 in CFI, 0.94 in TLI, and 0.00 in RMSEA. In older adults, the fitness of path model was very good with 1.00 in CFI, 1.15 in TLI, and 0.00 in RMSEA. This might imply if TMJ disorder level and number of infection is neglected in young aldults, mastication function will be significantly effected in older adults. Thus, development and management of health education program is needed based on self-assessment oral health level by young adults group and older adults group.

• Journal of Nutrition and Health
• /
• v.48 no.5
• /
• pp.441-450
• /
• 2015

Purpose: The purpose of this study was to investigate the effects of hypertension relevant nutrition knowledge on practices for prevention of hypertension in Chinese college students studying in South Korea. Methods: We conducted a survey of 276 Chinese students studying more than three months in Korean universities who were aged 19 and older about hypertension and sodium relevant nutrition knowledge, intake of salty processed food, hypertension prevention practices, and stages of behavior change for hypertension. Results: The average score on the questionnaire for hypertension relevant nutritional knowledge was 40.62 out of 50 points, and the average score for sodium relevant nutritional knowledge was 24 out of 30 points. Kimchi was the most frequently eaten salty processed food. The average score for hypertension prevention practices was 3.10 out of 5 points. The behavior change stages for prevention of hypertension were contemplation (47.1%), action (32.2%), and pre-contemplation (20.7%). The students received high scores on nutrition knowledge and showed significantly higher scores on the action stage than on pre-contemplation or contemplation. Nutritional knowledge of hypertension and sodium showed positive correlation with hypertension prevention practices, whereas negative correlation with salty processed food intake. Conclusion: Development of an education program for Chinese students in Korea on hypertension and sodium relevant nutritional knowledge is needed so that they can practice for prevention of hypertension.