• The Journal of Korean Association of Computer Education
• /
• v.8 no.3
• /
• pp.19-28
• /
• 2005

The purpose of the research is to examine design of the educational computing course for early childhood teacher. It was based on a development of the early childhood and unification of the computer activities in the curriculum. For these purpose, a number of literatures for the early childhood computer activities and teacher's want of the computer training were reviewed. In a result, it was made prototype of the educational computing course for early childhood teacher. Then, It was corrected and strengthened since then answered by 15 national kindergarten teacher located in Taejon. Based on these quality analysis, we designed a curriculum model for educational computing course for early childhood teacher. The developed the educational computing course for early childhood teacher would help the teachers to have critical attitude and necessary skills to access useful information using a computer and to appropriately respond to the changing educational environments by adopting computers in schools.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.5
• /
• pp.479-487
• /
• 2021

In this paper, we propose an Sound Event Detection (SED) model using self-training based on a noisy student model. The proposed SED model consists of two stages. In the first stage, a mean-teacher model based on an Residual Convolutional Recurrent Neural Network (RCRNN) is constructed to provide target labels regarding weakly labeled or unlabeled data. In the second stage, a self-training-based noisy student model is constructed by applying different noise types. That is, feature noises, such as time-frequency shift, mixup, SpecAugment, and dropout-based model noise are used here. In addition, a semi-supervised loss function is applied to train the noisy student model, which acts as label noise injection. The performance of the proposed SED model is evaluated on the validation set of the Detection and Classification of Acoustic Scenes and Events (DCASE) 2020 Challenge Task 4. The experiments show that the single model and ensemble model of the proposed SED based on the noisy student model improve F1-score by 4.6 % and 3.4 % compared to the top-ranked model in DCASE 2020 challenge Task 4, respectively.

• Journal of the Korean Society of Physical Medicine
• /
• v.18 no.2
• /
• pp.103-114
• /
• 2023

PURPOSE: This study compared the effects of computer-based and virtual reality-based cognitive rehabilitation programs on the cognitive function, upper limb function, activities of daily living, and their impact on the prefrontal cortex in convalescent stroke patients. METHODS: Ten recovering stroke patients were assessed for their cognitive function, upper limb function, and daily living activities using the Neurobehavioral Cognitive Status Examination, the Korean version of the Fugl-Meyer Assessment, and the Korean version of the Modified Barthel Index. The prefrontal cortex activity was measured with functional Near Infrared Spectroscopy. The virtual reality-based cognitive rehabilitation group utilized a program of daily living activities delivered via a laptop and Oculus Rift. The computer-based cognitive rehabilitation group performed various cognitive tasks on an all-in-one PC. Both groups underwent cognitive rehabilitation training for 30 minutes per day, three times a week, for six weeks, with identical conventional rehabilitation therapies in the hospital. RESULTS: Both programs positively impacted the cognitive and physical functions. On the other hand, the virtual reality-based cognitive rehabilitation program had a larger influence on improving the cognitive and physical functions of convalescing stroke patients. CONCLUSION: The virtual reality program suggests its potential to enhance cognitive and physical functions in convalescent stroke patients through increased engagement, focus, real-time feedback, and game elements, making it a promising rehabilitation approach.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.10
• /
• pp.157-165
• /
• 2021

Recently, studies using deep learning to analyze a large amount of text are being actively conducted. In particular, a pre-trained language model that applies the learning results of a large amount of text to the analysis of a specific domain text is attracting attention. Among various pre-trained language models, BERT(Bidirectional Encoder Representations from Transformers)-based model is the most widely used. Recently, research to improve the performance of analysis is being conducted through further pre-training using BERT's MLM(Masked Language Model). However, the traditional MLM has difficulties in clearly understands the meaning of sentences containing new words such as newly coined words. Therefore, in this study, we newly propose NTM(Newly coined words Target Masking), which performs masking only on new words. As a result of analyzing about 700,000 movie reviews of portal 'N' by applying the proposed methodology, it was confirmed that the proposed NTM showed superior performance in terms of accuracy of sensitivity analysis compared to the existing random masking.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.9
• /
• pp.1183-1189
• /
• 2021

There have been recently an increasing number of people working out at home. However, many of them do not have face-to-face guidance from experts, so they cannot effectively correct their wrong pose. This may lead to strain and injury to those doing home training. To tackle this problem, this paper proposes a video data-based pose classification and correction system for home training. The proposed system classifies poses using the multi-layer perceptron and pose estimation model, and corrects poses based on joint angels estimated. A voting algorithm that considers the results of successive frames is applied to improve the performance of the pose classification model. Multi-layer perceptron model for post classification shows the highest accuracy with 0.9. In addition, it is shown that the proposed voting algorithm improves the accuracy to 0.93.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.1
• /
• pp.27-35
• /
• 2021

In this paper, we propose a new stacking ensemble framework for deep learning models which reflects the distribution of label embeddings. Our ensemble framework consists of two phases: training the baseline deep learning classifier, and training the sub-classifiers based on the clustering results of label embeddings. Our framework aims to divide a multi-class classification problem into small sub-problems based on the clustering results. The clustering is conducted on the label embeddings obtained from the weight of the last layer of the baseline classifier. After clustering, sub-classifiers are constructed to classify the sub-classes in each cluster. From the experimental results, we found that the label embeddings well reflect the relationships between classification labels, and our ensemble framework can improve the classification performance on a CIFAR 100 dataset.

• Journal of Convergence for Information Technology
• /
• v.12 no.1
• /
• pp.31-38
• /
• 2022

This paper applies an expert independent unsupervised neural network learning-based multivariate time series data analysis model, MSCRED(Multi-Scale Convolutional Recurrent Encoder-Decoder), and to overcome the limitation, because the MCRED is based on Auto-encoder model, that train data must not to be contaminated, by using learning data sampling technique, called Subset Sampling Validation. By using the vibration data of power plant equipment that has been labeled, the classification performance of MSCRED is evaluated with the Anomaly Score in many cases, 1) the abnormal data is mixed with the training data 2) when the abnormal data is removed from the training data in case 1. Through this, this paper presents an expert-independent anomaly diagnosis framework that is strong against error data, and presents a concise and accurate solution in various fields of multivariate time series data.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.3
• /
• pp.43-54
• /
• 2024

Recently, there has been a lot of research on utilizing Language Models, and Large Language Models have achieved innovative results in various tasks. However, the practical application faces limitations due to the constrained resources and costs required to utilize Large Language Models. Consequently, there has been recent attention towards methods to effectively utilize models within given resources. Curriculum Learning, a methodology that categorizes training data according to difficulty and learns sequentially, has been attracting attention, but it has the limitation that the method of measuring difficulty is complex or not universal. Therefore, in this study, we propose a methodology based on data heterogeneity-based Curriculum Learning that measures the difficulty of data using reliable prior information and facilitates easy utilization across various tasks. To evaluate the performance of the proposed methodology, experiments were conducted using 5,000 specialized documents in the field of information communication technology and 4,917 documents in the field of healthcare. The results confirm that the proposed methodology outperforms traditional fine-tuning in terms of classification accuracy in both LoRA fine-tuning and full fine-tuning.

• Journal of information and communication convergence engineering
• /
• v.6 no.1
• /
• pp.100-104
• /
• 2008

Generally, it is difficult to find constant patterns on identifiers in a container image, since the identifiers are not normalized in color, size, and position, etc. and their shapes are damaged by external environmental factors. This paper distinguishes identifier areas from background noises and removes noises by using an ART2-based quantization method and general morphological information on the identifiers such as color, size, ratio of height to width, and a distance from other identifiers. Individual identifier is extracted by applying the 8-directional contour tracking method to each identifier area. This paper proposes a refined ART2-based RBF network and applies it to the recognition of identifiers. Through experiments with 300 container images, the proposed algorithm showed more improved accuracy of recognizing container identifiers than the others proposed previously, in spite of using shorter training time.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.31 no.11C
• /
• pp.1085-1091
• /
• 2006

The major goal of this paper is to develop a practically implemental radial basis function (RBF) neural network based multi-user detector (MUD) for direct sequence (DS)-CDMA system. This work is expected to provide an efficient solution for RBF based MUD by quickly setting up the proper number of RBF centers and their locations required in training. The basic idea in this research is to estimate all the possible RBF centers by using supervised ${\kappa-means$ clustering technique, and select the only centers which locate near seemingly decision boundary between centers, and reduce further by grouping the some of centers adjacent each other. Therefore, it reduces the computational burden for finding the proper number of RBF centers and their locations in the existing RBF based MUD, and ultimately, make its implementation practical.