• Journal of KIISE:Software and Applications
• /
• v.33 no.12
• /
• pp.1052-1061
• /
• 2006

This paper presents a method of using virtual examples to improve the classification accuracy for data with nominal attributes. Most of the previous researches on virtual examples focused on data with numeric attributes, and they used domain-specific knowledge to generate useful virtual examples for a particularly targeted learning algorithm. Instead of using domain-specific knowledge, our method samples virtual examples from a naive Bayesian network constructed from the given training set. A sampled example is considered useful if it contributes to the increment of the network's conditional likelihood when added to the training set. A set of useful virtual examples can be collected by repeating this process of sampling followed by evaluation. Experiments have shown that the virtual examples collected this way.can help various learning algorithms to derive classifiers of improved accuracy.

• Journal of Intelligence and Information Systems
• /
• v.11 no.1
• /
• pp.169-189
• /
• 2005

In order to derive a better classifier with a limited number of training examples, active teaming alternately repeats the querying stage fur category labeling and the subsequent learning stage fur rebuilding the calssifier with the newly expanded training set. To relieve the user from the burden of labeling, especially in an on-line environment, it is important to minimize the number of querying steps as well as the total number of query examples. We can derive a good classifier in a small number of querying steps by using only a small number of examples if we can select multiple of diverse, representative, and ambiguous examples to present to the user at each querying step. In this paper, we propose a cluster-based batch query selection method which can select diverse, representative, and highly ambiguous examples for efficient active learning. Experiments with various text data sets have shown that our method can derive a better classifier than other methods which only take into account the ambiguity as the criterion to select multiple query examples.

• Journal of KIISE:Software and Applications
• /
• v.31 no.8
• /
• pp.991-998
• /
• 2004

In many practical teaming problems including bioinformatics area, there is a small amount of labeled data along with a large pool of unlabeled data. Labeled examples are fairly expensive to obtain because they require human efforts. In contrast, unlabeled examples can be inexpensively gathered without an expert. A common method with unlabeled data for data classification and analysis is co-training. This method uses a small set of labeled examples to learn a classifier in two views. Then each classifier is applied to all unlabeled examples, and co-training detects the examples on which each classifier makes the most confident predictions. After some iterations, new classifiers are learned in training data and the number of labeled examples is increased. In this paper, we propose a new co-training strategy using unlabeled data. And we evaluate our method with two classifiers and two experimental data: WebKB and BIND XML data. Our experimentation shows that the proposed co-training technique effectively improves the classification accuracy when the number of labeled examples are very small.

• Journal of The Korean Association For Science Education
• /
• v.33 no.7
• /
• pp.1273-1284
• /
• 2013

The purpose of this study was to analyze the examples represented in school science textbooks by the function types for students' scientific conceptions. According to the framework of analysis, we selected lacking examples in the textbooks, and developed a questionnaire for students. The questionnaire was applied to 157 third grade students to survey their conceptions related to matter states and change of states. The ratio of students' scientific conceptions related to solid the state was high because distinct examples were represented in the textbook. However, the ratios of students' scientific conceptions related to the liquid and gas states were low because there were no distinct examples in the science textbook. Contrast examples and expansive examples of liquid and gas such as fog and steam need to be represented in science textbooks in order to help students construct scientific conceptions of matter states and change of states.

• Journal of The Korean Association For Science Education
• /
• v.28 no.6
• /
• pp.579-591
• /
• 2008

Recently, there have been studies about Typically- Perceived-Situations (TPSs) and about critical examples as a way to understand students' preconceptions with context. TPS is a situation arising immediately in one's mind when he or she thinks about the concept, while a critical example is an example that becomes the most helpful in learning the concept. We might explore how the context is involved in the process of students' conceptual understanding by examining TPSs and critical examples together. This study analyzed, through questionnaires and interviews, the characteristics of TPSs and those of critical examples that secondary students hold about 'force' and 'mechanical energy conversion.' Students' TPSs and critical examples showed different characteristics according to the concept. In a case of force that is related to everyday life, there were various situations as TPSs and critical examples. Unlike force, there were a few situations as TPSs and critical examples such as a falling ball in the case of mechanical energy conversion. Students tended to regard situations that are usually experienced and understood easily as TPSs or critical examples. On the basis of the results of this study, it is concluded that it would be a good strategy to teach science concepts for teachers to start with the TPS of a concept, to introduce the concept, and then to expose the attributes of the concept with critical examples.

• Journal of Fisheries and Marine Sciences Education
• /
• v.24 no.5
• /
• pp.699-711
• /
• 2012

An undergraduate course named computational structural analysis becomes more significant in recent years because of its important role in industries and the recent innovation in computer technology. Typically, the course consists of introduction to finite element method, utilization of general purpose finite element software, and examples focusing on static and linear analyses on various structural members such as a beam, truss, frame, arch, and cable. However, in addition to the static and linear analyses, current industries ask graduates to acquire basic knowledge on structural dynamics and nonlinear analysis, which are not listed in the conventional syllabus of the computational structural analysis. Therefore, this study develops geometrically nonlinear examples, which can help students to easily capture the fundamental nonlinear theory, software manipulation, and problem solving skills. For the purpose, five different examples are found, developed for the analyses of cables and cable nets, which naturally have strong geometrical non-linearity. In the paper, these examples are presented, discussed, and finally compared for a better subject development.

• Journal of The Korean Association For Science Education
• /
• v.14 no.2
• /
• pp.184-191
• /
• 1994

In this study, a part of high school science instructional materials is evaluated by Instructional Quality Profile(IQP) based on the Merrill's Component Display Theory(CDT). The CDT is based on the Gagne assumption of different conditions of learning for different outcomes. The IQP enables the user to check both the consistency and adequacy of existing cognitive instruction. The IQP can be used to predict student performance, and also to design and develop new insturctional materials. The instructional components are classified according to 5 task levels; An Use-Generalities on Newly Encountered Examples(UGeg), A Remember-Paraphrased -Generalities (RpG), A Remember-Verbatim-Generalities (RvG), A Remember-Paraphrased -Examples (Rpeg), A Remember-Verbatim-Examples(Rveg). And the instructional presentations are classified according to 4 levels: Explain Generalities(EG), Explain examples(Eeg), Inquiry Generalities(IG), Inquiry examples(Ieg). The instructional presentations are determined by instructional components of a related test item, and indexes of the presentation adequacy are calculated by the instructional presentations. The indexes of this study(0.17 - 0.44) were very low and it indicates that the instructional presentations were not adequate to the instructional components of the related text item.

• Journal of Korean Medical classics
• /
• v.23 no.2
• /
• pp.125-139
• /
• 2010

The character 'yeong(營)' has been used mixed with 'yeong(榮)' from the time of "Hwangjenaegyeong(黃帝內經)" up to now. One word can have a various meaning according to the context. So it is difficult to make a precise definition. Moreover as the words in Korean medicine are abstruse, it is necessary to classify and make the meaning straight with the words like 'yeong(營)' and 'yeong(榮)'. This study is focused on classifying the meanings and examples of 'yeong(營)' and 'yeong(榮)' by the dictionary definition, examples in medical classics, examples in "Hwangjenaegyeong(黃帝內經)". From this study, we get to know 'yeong(營)' and 'yeong(榮)' was used mixed with the concept of 'yeong-gi(營氣)' which means 'transporting nutrition'. The conclusion of this study is, from the dictionary definition and the aspect of oriental medicine physiology, using 'yeong(營)' is more reasonable than 'yeong(榮)' in both cases.

• International journal of advanced smart convergence
• /
• v.8 no.1
• /
• pp.141-146
• /
• 2019

These days, there are many applications using neural networks as parts of their system. On the other hand, adversarial examples have become an important issue concerining the security of neural networks. A classifier in neural networks can be fooled and make it miss-classified by adversarial examples. There are many research to encounter adversarial examples by using denoising methods. Some of them using GAN (Generative Adversarial Network) in order to remove adversarial noise from input images. By producing an image from generator network that is close enough to the original clean image, the adversarial examples effects can be reduced. However, there is a chance when adversarial noise can survive the approximation process because it is not like a normal noise. In this chance, we propose a research that utilizes high-level representation in the classifier by combining GAN network with a trained U-Net network. This approach focuses on minimizing the loss function on high representation terms, in order to minimize the difference between the high representation level of the clean data and the approximated output of the noisy data in the training dataset. Furthermore, the generated output is checked whether it shows minimum error compared to true label or not. U-Net network is trained with true label to make sure the generated output gives minimum error in the end. At last, the remaining adversarial noise that still exist after low-level approximation can be removed with the U-Net, because of the minimization on high representation terms.

• Proceedings of the IEEK Conference
• /
• 1999.06a
• /
• pp.997-1000
• /
• 1999

Conventional rule-based approaches have some problems caused by rule maintenance. Also they have some limitations to get the high quality translation results. This paper presents new English-Korean transfer approach that uses patterns and examples on limited domains. The use of patterns and examples can resolve the ambiguities and give high quality of MT Proposed approach can be applied in various NLP related area. Experimental results with a test corpus are discussed.