Earticle

본 연구는 여자핸드볼 경기에서 발생되는 움직임 정보를 바탕으로 앙상블 기법의 배깅과 부스팅 알고리즘의 예측력을 비교하고, 움직임 정보의 활용가능성을 분석하는데 목적이 있다. 연구의 목적을 달성하기 위하여 15번의 연습 경기에서 관성센서를 활용해 수집한 움직임 정보를 활용한 경기 결과예측을 랜덤포레스트와 Adaboost 알고리즘을 활 용해 비교·분석하였다. 연구결과 첫째, 랜덤포레스트 알고리즘의 예측률은 66.9 ± 0.1%로 나타났으며, Adaboost 알 고리즘의 예측률은 65.6 ± 1.6%로 나타났다. 둘째, 랜덤포레스트는 승리 결과는 모두 예측하였고, 패배의 결과는 하나 도 예측하지 못하였다. 반면, Adaboost 알고리즘은 승리 예측 91.4%, 패배예측 10.4%라고 나타났다. 셋째, 알고리즘 의 적합성 여부에서 랜덤포레스트는 과적합의 오류가 없었지만, Adaboost는 과적합의 오류가 나타났다. 본 연구결과를 바탕으로 스포츠경기를 예측할 때 움직임 정보도 활용 가능성을 확인하였으며, 랜덤포레스트 알고리즘이 보다 우수함을 확인하였다.

The purpose of this study is to compare the predictive power of the Bagging and Boosting algorithm of ensemble method based on the motion information that occurs in woman handball matches and to analyze the availability of motion information. To this end, this study analyzed the predictive power of the result of 15 practice matches based on inertial motion by analyzing the predictive power of Random Forest and Adaboost algorithms. The results of the study are as follows. First, the prediction rate of the Random Forest algorithm was 66.9 ± 0.1%, and the prediction rate of the Adaboost algorithm was 65.6 ± 1.6%. Second, Random Forest predicted all of the winning results, but none of the losing results. On the other hand, the Adaboost algorithm shows 91.4% prediction of winning and 10.4% prediction of losing. Third, in the verification of the suitability of the algorithm, the Random Forest had no overfitting error, but Adaboost showed an overfitting error. Based on the results of this study, the availability of motion information is high when predicting sports events, and it was confirmed that the Random Forest algorithm was superior to the Adaboost algorithm.

High-risk pregnancy endangers both mother and baby, with one maternal death every two minutes in 2023. This study has proposed three Gradient Boosting models?GB-Base, GB-SMOTE, and ZOA-GB?using the West Lombok Pregnancy Risk Prediction Dataset. GB-Base and GB-SMOTE have achieved 90.46% and 90.28% accuracy, while ZOA-GB, using 10 selected features, has reached 88.89%. GB-SMOTE has shown the best performance with an F-score of 84.41%. SHAP has identified Maternal Age, Hemoglobin, and Parity as key features, and DiCE has validated feature-driven prediction control. The study is limited by a single-source dataset, the absence of external-validation, and unexplored optimizers.

Background: Genotype imputation is an important process of predicting unknown genotypes, which uses reference population with dense genotypes to predict missing genotypes for both human and animal genetic variations at a low cost. Machine learning methods specially boosting methods have been used in genetic studies to explore the underlying genetic profile of disease and build models capable of predicting missing values of a marker. Methods: In this study strategies and factors affecting the imputation accuracy of parent-offspring trios compared from lower-density SNP panels (5 K) to high density (10 K) SNP panel using three different Boosting methods namely TotalBoost (TB), LogitBoost (LB) and AdaBoost (AB). The methods employed using simulated data to impute the un-typed SNPs in parent-offspring trios. Four different datasets of G1 (100 trios with 5 k SNPs), G2 (100 trios with 10 k SNPs), G3 (500 trios with 5 k SNPs), and G4 (500 trio with 10 k SNPs) were simulated. In four datasets all parents were genotyped completely, and offspring genotyped with a lower density panel. Results: Comparison of the three methods for imputation showed that the LB outperformed AB and TB for imputation accuracy. The time of computation were different between methods. The AB was the fastest algorithm. The higher SNP densities resulted the increase of the accuracy of imputation. Larger trios (i.e. 500) was better for performance of LB and TB. Conclusions: The conclusion is that the three methods do well in terms of imputation accuracy also the dense chip is recommended for imputation of parent-offspring trios.

NGBoost has shown promising results in probabilistic and point estimation tasks. However, it is vague still whether this method can be scalable to neural architecture system since its base learner is based on decision trees. To resolve this, we design a Neural-NGBoost framework by replacing the base learner with lightweight neural networks and introducing joint gradient estimation for boosting procedure. Based on natural gradient boosting, we iteratively update the neural based learner by inferring natural gradient and update the parameter score with its probabilistic distribution. Experimental results show Neural-NGBoost achieves superior performance across various datasets compared to other boosting methods.

The increasing demand for high-speed, low-latency applications, especially with 5G mmWave technology, has led to challenges in TCP performance due to signal blockages, small buffers, and high Packet Error Rates (PERs). Existing congestion control algorithms (CCAs) struggle to fully utilize available bandwidth under these conditions. This paper proposes MSS-TCP, a novel congestion control algorithm designed for mmWave networks. MSS-TCP dynamically adjusts the congestion window (cwnd) based on the maximum segment size (MSS) and round-trip time (RTT), improving bandwidth utilization and congestion adaptability. The simulation results using the ns-3 network simulator show that MSS-TCP outperforms state-of-the-art CCAs, including NewReno, HighSpeed, CUBIC, and Bottleneck Bandwidth and Round-trip propagation time (BBR), and Fuzzy Logic-based (FB-TCP), particularly when the buffer matches the bandwidth-delay product (BDP), achieving a 24.26% to 45.43% improvement in throughput compared to BBR while maintaining low latency. These findings demonstrate that MSS-TCP enhances TCP performance in 5G mmWave networks, making it a promising solution for next-generation wireless communication.

.

본 논문에서는 복잡한 배경에서의 사람의 머리 추적에 있어서 효과적인 Adaptive Boosting에 의한 방법을 제 안한다. 하나의 특징 추출 방법은 사람의 머리를 모델링하기에는 부족하다. 따라서 본 연구에서는 여러 가지 특징 추 출 방법을 병행하여 정확한 머리 검출을 시도하였다. 머리 영상의 특징 추출은 sub-region과 Haar 웨이블릿 변환(Haar wavelet transform)을 이용하였다. Sub-region은 머리의 지역적인 특징을 나타내고, Haar 웨이블릿 변환은 얼굴의 주파 수 특성을 나타내기 때문에 이들을 이용하여 특징을 추출하면 효과적인 모델링이 가능해 진다. 실시간으로 입력되는 영상에서 사람의 머리를 추적하기 위하여 제안하는 방법에서는 3가지 형태의 Harr-wavelet 특징을 AdaBoosting 알고 리즘으로 학습한 후 결과를 이용하였다. 원래 AdaBoosting 알고리즘은 학습시간이 매우 길며 학습데이터가 변하면 다 시 학습을 수행해야 하는 단점이 존재한다. 이 단점을 극복하기 위하여 제안하는 방법에서는 캐스케이드를 이용한 AdaBoosting의 효율적인 학습방법을 제안한다. 이 방법은 머리 영상에 대한 학습시간은 감소시키며, 학습데이터의 변 화에도 효율적으로 대처할 수 있다. 이 방법은 학습과정을 레벨별로 분리한 후 중요도가 높은 학습데이터를 다음 단 계에 반복적으로 적용시킨다. 제안하는 방법이 적은 학습 시간과 학습 데이터를 사용해서 우수한 성능을 가지는 분류 기를 생성하였다. 또한, 이 방법은 다양한 머리데이터를 가진 실시간 영상데이터에 적용한 결과 다양한 머리를 정확 하게 검출 및 추적하였다.

This paper proposes an effective method using Adaptive Boosting to track a person's head in complex background. By only one way to feature extraction methods are not sufficient for modeling a person's head. Therefore, the method proposed in this paper, several feature extraction methods for the accuracy of the detection head running at the same time. Feature Extraction for the imaging of the head was extracted using sub-region and Haar wavelet transform. Sub-region represents the local characteristics of the head, Haar wavelet transform can indicate the frequency characteristics of face. Therefore, if we use them to extract the features of face, effective modeling is possible. In the proposed method to track down the man's head from the input video in real time, we ues the results after learning Harr-wavelet characteristics of the three types using AdaBoosting algorithm. Originally the AdaBoosting algorithm, there is a very long learning time, if learning data was changes, and then it is need to be performed learning again. In order to overcome this shortcoming, in this research propose efficient method using cascade AdaBoosting. This method reduces the learning time for the imaging of the head, and can respond effectively to changes in the learning data. The proposed method generated classifier with excellent performance using less learning time and learning data. In addition, this method accurately detect and track head of person from a variety of head data in real-time video images.

SUMMARY I. 서 론 II. 재료 및 방법 III. 결 과 IV. 고 찰 V. 적 요 VI. 인용문헌

This study focuses on students’ journal writing with the attempt to promote students’ autonomy in writing in English as well as their writing accuracy and fluency. About 115 students from four faculties participated in the research. Students were asked to keep their journals in a systematic way. Students were constantly given guidance as they continually worked through their own writing. They were encouraged to reflect on their writing by learning to correct their own mistakes. Furthermore, the students were motivated to set goals in order to improve their subsequent journal writing. Moreover, both fluency and accuracy were encouraged during the entire journal writing process. With the help of new technology and a little support from the teachers, these students emerge with their autonomy to write in English. This experience will also generate their life-long learning and writing in English as well as establishing positive attitude towards studying English on their own in the future.

This paper proposes a novel geometric mean (GM) optimization-based boosting algorithm (GMOPTBoost) to improve the performance of boosting ensembles applied to solve the class imbalance problem in bankruptcy prediction. GMOPTBoost derives the best prediction by applying Gaussian gradient descent method to find the set of weights assigned to base classifiers to optimize GM. The main findings are as follows. First, the class imbalance problem has a negative effect on the performance. As IR values increase, the performances of boosting ensembles decreases. Second, GMOPTBoost makes a significant contribution to performance improvements of AdaBoost ensembles trained on imbalanced datasets.

Prediction of blood glucose (BG) values in type 1 diabetes (T1D) remains an essential and challenging issue. Recently, machine learning methods have been used to solve this problem. We present a forecasting model based on extreme gradient boosting (XGBoost) regression to estimate the BG value of T1D patients in this study. We developed the models using clinical datasets from five T1D patients and forecasted the BG value for the following prediction horizons (PHs) of 10, and 20 minutes. Datasets are divided in two parts, around 60:40 for training and testing. We compared the performance of our proposed model to existing models using several performance metrics, including root mean square error (RMSE), mean absolute percentage error (MAPE), and coefficient of determination (R2). Our suggested XGBoost model outperformed existing models, with average RMSE, MAPE, and R2 of 12.57 mg/dL, 7.15%, 0.94, and 21.93 mg/dL, 13.72%, 0.84 for PH of 10 and 20 minutes, respectively. Finally, the findings of this study are intended to be applied to improve diabetes care.

Background: Chrysanthemum zawadskii Herbich var. latilobum Kitamura extracts (CZE) is used to treat of various inflammatory and chronic diseases. Although commonly adminstered, no reports suggest its potential usage as an immune-booster. Methods and Results: In the cyclophosphamide (CP, 5 ㎎/㎏)-induced immunosuppression model, the immune-boosting effect of CZE (oral administration for 28 days with CP) was evalu- ated by investigating cell proliferation, activity of natural killer (NK) cells, cytotoxic T lympho- cytes, and production of CP-repressed cytokines [such as tumor necrosis factor-α, interferon-γ, interluekin (IL)-2, and IL-12] in isolated splenocytes. In Vitro, CZE treatment enhanced cell prolif- eration, activity of NK cells and cytotoxic T lymphocytes activity, and production of CP-repressed cytokines. In Vivo, CZE treatment promoted the production of white blood cells, lymphocytes, medium-sized cells, and granulocytes and proliferation of NK cells in CP-induced immunosup- pressed mice. Additionally, the CZE treatment restored the TNF-α and IL-12 reduced by CP to nor- mal levels and prevented spleen tissue damage. Conclusions: CZE could be an effective immune-booster and potentially be added to functional foods to enhance immunity.

이 연구의 목적은 학교 독서교육의 활성화 방안을 모색하기 위함이다. 연구 대상은 경남지역의 학교도서관을생각하는사람들의모임(이하 학생사모)의 독서교육 연구·실천 활동이며, 연구방법은 사례연구이다. 이를 위한 데이터 수집은 학생사모 회원을 대상으로 반구조화된 설문지를 통한 인터뷰, 관련 문서, 홈페이지자료 등을 활용하였다. 학생사모의 활동을 분석한 결과 학생사모가 이룬 학교 독서교육의 성과와 전망을 밝혔다. 향후 학교 독서교육 활성화 방안으로 학교 독서교육은 반드시 학교도서관과 연계되어야 하며, 학교 내·외의 조직과 연대와 협력의 필요성을 제시하였다.

The purpose of this study is to explore ways for boosting reading education at school. Research object is research and practice activities of 'Society for School Library' at Gyeongnam area. Research method is case study. Data for this study were collected by interview with 'Society for school library' members through half structured survey, their documents and homepage site database. The results of analyzing the activities of 'Society for School Library' were explored reading education's outcomes and prospects that performed impassionedly by them. It suggests ways for boosting of reading education at school that under any circumstances reading education at school must connect with school library and the necessity of ties and collaborations with organizations inside and outside of school.

본 연구는 경제활동 참여에 따른 장애인의 삶의 만족 예측 요인을 분석하고, 취업 장애인과 비경제활동 장애인 간 공통점과 차이점을 비교하였다. 기존 연구의 선형적 통계 방법을 벗어나 기계학습 기반의 앙상블 기법과 SHAP 분석을 활용하여 변수의 예측 중요도와 비선형적 구조를 분석하였다. 분석 결과, 두 집단 모두에서 삶의 만족을 결정하는 보편적인 영역은 심리·사회적 요인이었으며, 특히, ‘행복 정도’와 ‘장애 수용’, ‘스트레스’ 등이 핵심 변수로 나타났다. 취업 장애인의 경우는 ‘일상생활 및 사회활동’과 ‘취업 및 고용’ 요인의 기여도가 높았고, ‘사회활동 참여 정도’와 ‘직무만족도’ 등이 주요 핵심 변수였다. 비경제활동 장애인은 경제 요인의 영향이 두드러졌고, ‘가구의 경제적 어려움’과 ‘한 달 평균 생활비 총계’ 등이 주요 변수로 나타났다. 나아가 SHAP 분석을 통해 주요 변수들이 삶의 만족에 미치는 비선형적 특성이 확인되었다. 취업 장애인 집단에서는 ‘코로나 변화 사항_시설 이용’, ‘가구의 경제적 어려움’ 등이, 비경제활동 장애인 집단에서는 ‘직업능력’과 ‘한 달 평균 생활비 총계’ 등이 복합적인 패턴을 보였다. 이러한 결과를 토대로 경제활동 참여에 따른 장애인의 삶의 만족 향상을 위한 개별 전략과 정책적 함의를 논의하였다.

This study analyzed factors predicting life satisfaction among people with disabilities according to their economic participation and compared similarities and differences between employed and non-economically active individuals with disabilities. Moving beyond the linear statistical methods used in previous studies, this research employed machine learning?based ensemble techniques and SHAP analysis to examine the predictive importance and non-linear effects of key variables. The results showed that psychosocial factors were common determinants of life satisfaction in both groups, with level of happiness, acceptance of disability, and stress identified as key variables. Among employed people with disabilities, life satisfaction was strongly associated with daily life and social activity factors as well as employment-related factors, particularly participation in social activities and job satisfaction. In contrast, the life satisfaction of non-economically active people with disabilities was more strongly influenced by economic factors, with household financial difficulties and average monthly living expenses emerging as major predictors. Furthermore, SHAP analysis revealed complex non-linear patterns in the effects of several variables. For the employed group, these patterns were evident in COVID-19?related changes in facility use and household financial difficulties. For the non-economically active group, vocational competence and average monthly living expenses exhibited pronounced non-linear effects. Based on these findings, this study discusses individualized strategies and policy implications for improving life satisfaction among people with disabilities in relation to economic participation.

전이 학습은 이미 학습된 딥러닝 모델을 초기 모델로 활용하여, 다른 데이터에서 높은 성능 을 발휘하는 기술이다. 특히, 학습에 사용할 데이터의 질과 양이 충분하지 않을 때 전이 학습은 매우 유용한 것으로 알려져 있기에 높은 성능 안정성과 많은 데이터를 필요로 하는 자율주행 차 인식 분야에 응용될 수 있다. 그러나 많은 선행 연구들은 전이 학습 알고리즘 자체의 성능 향상에 초점을 맞추었다. 본 연구는 기존의 알고리즘 중심 접근 방식에서 벗어나, 여러 데이터 셋의 특징 추출기 출력을 융합한 심층 모델 융합 기법을 전이 학습에 적용하고자 한다. 실험 결과, 이러한 심층 모델 융합 기법이 전이 학습의 성능을 향상시킬 수 있음을 확인했다. 이 결 과는 앞으로 데이터가 부족한 자율주행 자동차 분야에서 전이 학습에 활용되어 물체 인지 성 능의 향상을 달성할 수 있을 것으로 기대된다.

Transfer learning is a technique that leverages a deep learning model trained on a specific dataset as an initial model that allows fast training of a high-performing model on another dataset. Because a pre-trained model already learns how to extract the features from previously trained data, it allows for faster and better performance on new datasets compared to models that are initialized randomly. Transfer learning is particularly useful when the quality and quantity of the data to be learned are insufficient. On the other hand, most studies focused on improving the performance of transfer learning algorithms themselves. This study departs from the existing algorithm-centric research approach and aims to incorporate deep model fusion techniques that combine the outputs of feature extractors from different datasets into transfer learning. These experiments show that the application of deep model fusion improves the performance of transfer learning. These findings will be applicable to transfer learning in various domains with limited data.

인터넷 사용의 증가로 인한 외부위협에 대응하는 방안의 하나로서 네트워크 침입탐지를 위한 머신러닝 방식의 연구가 활발하다. 그런데 침입탐지를 위한 머신러닝 알고리즘의 적용에 있어서 소수 클래스로 인해 발생하는 데이터 불균형 문제가 지적된다. 네트워크 침입탐지를 비롯한 일반적인 분류에서는 데이터 불균형에 의한 문제점보다는 대개 전체 모형의 정확도가 목표인 경우가 많아서, 데이터 불균형에 대한 대처가 쉽지 않을 수 있다. 네트워크 침입탐지에 대한 랜덤포레스트(RF) 모형에서 데이터 불균형 문제가 있음을 본 논문에서 확인하 고, 이를 개선하기 위한 그래디언트부스팅 분류기(GBC)의 구성 및 효과를 정리하였다. 랜덤포레스트 모형과 그래디언트부스팅 분류기의 구성은 KDDTrain+ 데이터를 활용하였고, KDDTest+ 데이터로써 구성한 모형을 평가하였다. RF모형과 GBC의 성능에서 보이는 차이점 은 빈도가 낮은 침입유형에서 RF모형에 비해 정확도의 큰 변화 없이 GBC의 정밀도 및 재현율이 높은 것을 확인하였다. GBC의 이러한 효과는 특히 피해가 큰 침입유형에 대해 높은 탐지율로써 전체 피해를 줄이는 효과를 기대할 수 있다.

As a way to respond to external threats due to the increase of Internet usage, researches on machine learning methods for network intrusion detection becomes more active. However the problem of data imbalance caused by minority classes is pointed out in the application of machine learning algorithms for intrusion detection. In general classification problems including network intrusion detection, the accuracy of the entire model is often the goal rather than handling the problems caused by such minority classes, so it may not be easy to deal with data imbalance. In this paper, we checked that there is a data imbalance problem in the random forest model used in network intrusion detection, and organized the composition and effect of the gradient boosting classifier for this point. The Random Forest (RF) model and the Gradient Boosting Classifier (GBC) were constructed using the KDDTrain+ data and evaluated using the KDDTest+ data. The difference in the performance of the RF model and the GBC is that the precision and recall of the GBC are higher than that of the RF model without a significant change in accuracy in low-frequency intrusion types. This effect of GBC is expected to have the effect of reducing the overall damage by detecting intrusion types that cause particularly serious damage with a higher probability.