• The Korean Journal of Applied Statistics
• /
• v.37 no.5
• /
• pp.615-629
• /
• 2024

In the field of big data mining, the challenge of imbalanced classification problem has been actively researched for decades. While imbalanced data issues manifest in various forms, past research mainly focused on addressing sample size imbalance between classes. However, recent studies have revealed that rather than the imbalance in sample size alone, the degradation of classification performance significantly worsens when the class overlap is combined. In response, this study introduces GC-Forest (GIR-based canonical forest), an effective ensemble classification method that utilizes weighted resampling technique considering the degrees of overlap between classes. This method measures the imbalance ratio in terms of class overlap at each stage of ensemble and balances the classes by increasing the representativeness of the minority class. Additionally, to improve overall classification performance, the GC-Forest method adopts the canonical forest method as an ensemble classifier, which is designed to enhance both the performance and diversity of individual classifiers. The performance of the proposed method was compared and verified through experiments using 14 different types of real imbalanced data. GC-Forest showed very competitive classification performance in terms of AUC, PR-AUC, G-mean, and F1-score compared to 7 other ensemble methods.

• Journal of the Korean Society of Earth Science Education
• /
• v.16 no.2
• /
• pp.182-195
• /
• 2023

The purpose of this study was to analyze reports from research middle schools based on the criteria for process-focused assessment to find out how the characteristics of process-focused assessment were being implemented in middle school science classes. The analysis criteria for the characteristics of process-focused assessment (integration of lessons and assessments, evaluation elements and methods, content and timing of feedback, and learner growth management) were extracted. Using the analysis framework, the result reports of seven research middle schools for process-focused assessment were analyzed. In terms of integration of lessons and assessments, when the process-focused assessment was operated, the class and evaluation plan were well implemented based on the curriculum achievement standards, but the process-focused assessment was recognized as a performance evaluation. In terms of evaluation elements and methods, the evaluation element for knowledge was the main component, and competency was presented in the planning stage, but competency was not dealt with in class execution. The evaluation method was biased toward teacher-centered observation evaluation and written test, and the setting of scoring criteria for each evaluation element was insufficient. In terms of the content and timing of feedback, feedback was mainly provided based on achievement confirmation, but no case was found in which scaffolding was provided at an appropriate time for insufficient parts in the learning process. In terms of the learner's growth management, the competencies cultivated through science classes were included in the detailed subject specialties of the school record. However, little was shown in the report on how to systematically manage the process of developing learners' competencies and reflect the evaluation results to teachers' class improvement.

• Korean Journal of Radiology
• /
• v.22 no.2
• /
• pp.281-290
• /
• 2021

Objective: To assess the performance of content-based image retrieval (CBIR) of chest CT for diffuse interstitial lung disease (DILD). Materials and Methods: The database was comprised by 246 pairs of chest CTs (initial and follow-up CTs within two years) from 246 patients with usual interstitial pneumonia (UIP, n = 100), nonspecific interstitial pneumonia (NSIP, n = 101), and cryptogenic organic pneumonia (COP, n = 45). Sixty cases (30-UIP, 20-NSIP, and 10-COP) were selected as the queries. The CBIR retrieved five similar CTs as a query from the database by comparing six image patterns (honeycombing, reticular opacity, emphysema, ground-glass opacity, consolidation and normal lung) of DILD, which were automatically quantified and classified by a convolutional neural network. We assessed the rates of retrieving the same pairs of query CTs, and the number of CTs with the same disease class as query CTs in top 1-5 retrievals. Chest radiologists evaluated the similarity between retrieved CTs and queries using a 5-scale grading system (5-almost identical; 4-same disease; 3-likelihood of same disease is half; 2-likely different; and 1-different disease). Results: The rate of retrieving the same pairs of query CTs in top 1 retrieval was 61.7% (37/60) and in top 1-5 retrievals was 81.7% (49/60). The CBIR retrieved the same pairs of query CTs more in UIP compared to NSIP and COP (p = 0.008 and 0.002). On average, it retrieved 4.17 of five similar CTs from the same disease class. Radiologists rated 71.3% to 73.0% of the retrieved CTs with a similarity score of 4 or 5. Conclusion: The proposed CBIR system showed good performance for retrieving chest CTs showing similar patterns for DILD.

• International Journal of Control, Automation, and Systems
• /
• v.1 no.1
• /
• pp.11-18
• /
• 2003

The parametric approaches to sliding mode design are newly proposed for the class of multivariable systems. Our approach is based on an explicit formula for representing all the slid-ing modes using the Lyapunov matrices of full order. By manipulating Lyapunov matrices, the sliding modes which satisfy the design criteria such as the quadratic performance optimization and robust stability to parametric uncertainty, etc., can be easily obtained. The proposed ap-proach enables us to adopt a variety of Lyapunov- (or Riccati-) based approaches to the sliding mode design. Applications to the quadratic performance optimization problem, uncertain systems, systems with uncertain state delay, and the pole-clustering problem are discussed.

• ETRI Journal
• /
• v.33 no.3
• /
• pp.458-461
• /
• 2011

With the introduction of the new generation RFID technology, EPCglobal Class-1 Generation-2, there is considerable interest in improving the performance of the framed slotted Aloha (FSA)-based tag collision arbitration protocol. We suggest a novel time-optimal anti-collision algorithm for the FSA protocol. Our performance evaluation demonstrates that our algorithm outperforms other tag collision arbitration schemes.

• Earthquakes and Structures
• /
• v.25 no.3
• /
• pp.199-208
• /
• 2023

Seismic performance analysis is one of the fundamental steps in the design of new or retrofitting buildings. In the seismic performance analysis, the adapted spectral acceleration curve for a given site mainly governs the seismic behavior of buildings. Since every soil site (class) has a different impact on the spectral accelerations of input motions, different spectral acceleration curves have to be involved for every soil class that the building is located on top of. Modern seismic design codes (e.g., Eurocode 8, EC8, or Turkish Building Earthquake Code, TBEC) provide design response spectra for all the soil classes to be used in the building design or retrofitting. This research aims to evaluate the EC8 and TBEC based design response spectra using the spectra of real earthquake input motions that occurred (and were recorded at only soil classes A, B and C, no recording is available at soil class D) in a specific area in Turkey. It also conducts response spectrum analyses of 5, 10 and 13 floor reinforced concrete building models under EC8, TBEC and actual spectral response curves. The results indicate that the EC8 and especially TBEC given design response spectra cannot be able to represent the mean actual spectral acceleration curves at soil classes A, B and C. This is particularly observed at periods higher than 0.3 s, 0.42 s and 0.55 s for the TBEC design response spectra, 0.54 s, 0.65 s and 0.84 s for the EC8 design response spectra at soil classes A, B and C, respectively. This is also reflected to the shear forces of three building models, as actual spectral acceleration curves lead to the highest shear forces, followed by the shear forces obtained from EC8 and, then, the TBEC design response spectra.

• Journal of Electrical Engineering and Technology
• /
• v.12 no.2
• /
• pp.632-642
• /
• 2017

This paper presents an improved approach towards reducing the switching loss of insulated gate bipolar transistors (IGBTs) for a medium-capacity-class power conditioning system (PCS). In order to improve the switching performance, the switching operation is analyzed, and based on this analysis, an improved switching method that reduces the switching time and switching loss is proposed. Compared to a conventional gate drive scheme, the switching loss, switching time, and delay are improved in the proposed gate driving method. The performance of the proposed gate driving method is verified through several experiments.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.18 no.6
• /
• pp.285-292
• /
• 2023

With the wide spread of smart farms and the advancements in IoT technology, it is easy to obtain additional data in addition to crop images. Consequently, deep learning-based crop disease diagnosis research utilizing multimodal data has become important. This study proposes a crop disease diagnosis method using multimodal supervised contrastive learning by expanding upon the multimodal self-supervised learning. RandAugment method was used to augment crop image and time series of environment data. These augmented data passed through encoder and projection head for each modality, yielding low-dimensional features. Subsequently, the proposed multimodal supervised contrastive loss helped features from the same class get closer while pushing apart those from different classes. Following this, the pretrained model was fine-tuned for crop disease diagnosis. The visualization of t-SNE result and comparative assessments of crop disease diagnosis performance substantiate that the proposed method has superior performance than multimodal self-supervised learning.

• International Journal of Computer Science & Network Security
• /
• v.22 no.10
• /
• pp.406-412
• /
• 2022

Scene classification of very high-resolution (VHR) imagery can attribute semantics to land cover in a variety of domains. Real-world application requirements have not been addressed by conventional techniques for remote sensing image classification. Recent research has demonstrated that deep convolutional neural networks (CNNs) are effective at extracting features due to their strong feature extraction capabilities. In order to improve classification performance, these approaches rely primarily on semantic information. Since the abstract and global semantic information makes it difficult for the network to correctly classify scene images with similar structures and high interclass similarity, it achieves a low classification accuracy. We propose a VHR remote sensing image classification model that uses extracts the global feature from the original VHR image using an EfficientNet-V2L CNN pre-trained to detect similar classes. The image is then classified using a multilayer perceptron (MLP). This method was evaluated using two benchmark remote sensing datasets: the 21-class UC Merced, and the 38-class PatternNet. As compared to other state-of-the-art models, the proposed model significantly improves performance.

• Journal of Fashion Business
• /
• v.27 no.5
• /
• pp.121-134
• /
• 2023

With the development of information and communication technology, interest in new educational approaches that can enhance the learning performance of learners with improved information literacy skills is increasing, and universities are actively promoting educational innovation to foster the talents required by society. In the field of fashion studies education, which is closely related to the fashion industry, there is a strong need to develop field-linked educational programs that reflect the trends in the industry and changes in the educational system. The purpose of this study was to introduce industry-coupled problem-based learning (IC-PBL) to the course "Understanding Fashion Consumption Trends" for non-fashion majors to reflect the current needs and strengthen the educational effectiveness of the learners through a survey. A seven-step curriculum (introduction to the class, practitioner's problem, learner's problem analysis, organizing concepts related to variables, information collection and scenario writing, presentation and scenario proposal, and evaluation) not only enhanced learners' understanding of fashion consumption trends and the fashion industry but also greatly amplified learners' satisfaction with the class. The results of the survey showed that the seven-step curriculum was effective in increasing learners' self-directed learning ability, problem-solving ability, and confidence in learning. Self-directed learning ability was stronger than other factors, consistent with the core principle of problem-based learning to empower learners to take the initiative and promote self-directed learning. Each factor analyzed was positively correlated.