• Journal of Internet Computing and Services
• /
• v.25 no.4
• /
• pp.7-21
• /
• 2024

There are various machine learning techniques such as Reinforcement Learning, Deep Learning, Neural Network Learning, and so on. In recent, Large Language Models (LLMs) are popularly used for Generative AI based on Reinforcement Learning. It makes decisions with the most optimal rewards through the fine tuning process in a particular situation. Unfortunately, LLMs can not provide any explanation for how they reach the goal because the training is based on learning of black-box AI. Reinforcement Learning as black-box AI is based on graph-evolving structure for deriving enhanced solution through adjustment by human feedback or reinforced data. In this research, for mutually exclusive decision-making, Mutually Exclusive Learning (MEL) is proposed to provide explanations of the chosen goals that are achieved by a decision on both ends with specified conditions. In MEL, decision-making process is based on the tree-based structure that can provide processes of pruning branches that are used as explanations of how to achieve the goals. The goal can be reached by trade-off among mutually exclusive alternatives according to the specific contextual conditions. Therefore, the tree-based structure is adopted to provide feasible solutions with the explanations based on the pruning branches. The sequence of pruning processes can be used to provide the explanations of the inferences and ways to reach the goals, as Explainable AI (XAI). The learning process is based on the pruning branches according to the multi-dimensional contextual conditions. To deep-dive the search, they are composed of time window to determine the temporal perspective, depth of phases for lookahead and decision criteria to prune branches. The goal depends on the policy of the pruning branches, which can be dynamically changed by configured situation with the specific multi-dimensional contextual conditions at a particular moment. The explanation is represented by the chosen episode among the decision alternatives according to configured situations. In this research, MEL adopts the tree-based learning model to provide explanation for the goal derived with specific conditions. Therefore, as an example of mutually exclusive problems, employment process is proposed to demonstrate the decision-making process of how to reach the goal and explanation by the pruning branches. Finally, further study is discussed to verify the effectiveness of MEL with experiments.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.31 no.6
• /
• pp.1227-1236
• /
• 2021

Adversarial attack, one of the attacks on deep learning classification model, is attack that add indistinguishable perturbations to input data and cause deep learning classification model to misclassify the input data. There are various adversarial attack algorithms. Accordingly, many studies have been conducted to detect adversarial attack but few studies have been conducted to classify what adversarial attack algorithms to generate adversarial input. if adversarial attacks can be classified, more robust deep learning classification model can be established by analyzing differences between attacks. In this paper, we proposed a model that detects and classifies adversarial attacks by constructing a random forest classification model with input features extracted from a target deep learning model. In feature extraction, feature is extracted from a output value of hidden layer based on class predicted by the target deep learning model. Through Experiments the model proposed has shown 3.02% accuracy on clean data, 0.80% accuracy on adversarial data higher than the result of pre-existing studies and classify new adversarial attack that was not classified in pre-existing studies.

• ETRI Journal
• /
• v.32 no.5
• /
• pp.766-773
• /
• 2010

Support vector machine (SVM) active learning plays a key role in the interactive content-based image retrieval (CBIR) community. However, the regular SVM active learning is challenged by what we call "the small example problem" and "the asymmetric distribution problem." This paper attempts to integrate the merits of semi-supervised learning, ensemble learning, and active learning into the interactive CBIR. Concretely, unlabeled images are exploited to facilitate boosting by helping augment the diversity among base SVM classifiers, and then the learned ensemble model is used to identify the most informative images for active learning. In particular, a bias-weighting mechanism is developed to guide the ensemble model to pay more attention on positive images than negative images. Experiments on 5000 Corel images show that the proposed method yields better retrieval performance by an amount of 0.16 in mean average precision compared to regular SVM active learning, which is more effective than some existing improved variants of SVM active learning.

• Journal of Korea Multimedia Society
• /
• v.10 no.12
• /
• pp.1679-1686
• /
• 2007

In this paper, we introduce the sharable and reusable learning objects which are suitable for XML Web services in e-learning systems. These objects are extracted from the principles of pedagogical paradigms for reusable learning units. We call them LIO (Learning Item Object) objects. Existing models, such as Web-hosted and ASP-oriented service model, are difficult to cooperate and integrate among the different kinds of e-learning systems. So we developed the LIO objects that are suitable for XML Web services. The reusable units that are extracted from pedagogical paradigms are tutorial item, resource, case example, simulation, problems, test, discovery and discussion. And these units correspond to the LIO objects in our learning object model. As a result, the proposed model is that learner and instruction designer should increase the power of understanding about learning contents that are based on pedagogical paradigms. By using XML Web services, this guarantees the integration and interoperation of the different kinds of e-learning systems in distributed environments and so educational organizations can expect the cost reduction in constructing e-learning systems.

• Journal of Korea Multimedia Society
• /
• v.6 no.4
• /
• pp.676-684
• /
• 2003

eLearning can be defined as an approach to teaching and teaming that utilises Internet technologies to communicate and collaborate in an educational context. This includes technology that supplements traditional classroom training with web-based components and learning environments where the educational process is experienced online. The use of hypertext as an educational tool has a very rich history. The advent of the internet and one of its major application, the world wide web (WWW), has given a tremendous boost to the theory and practice of hypermedia systems for educational purposes. However, the web suffers from an inability to satisfy the heterogeneous needs of a large number of users. For example, web-based courses present the same static teaming material to students with widely differing knowledge of the subject. Adaptive hypermedia techniques can be used to improve the adaptability of eLearning. In this paper we report an approach to the design a unified implementation framework suitable for web-based eLearning that accommodates the three main dimensions of hypermedia adaptation: content, navigation, and presentation. The framework externalises the adaptation strategies using XML notation. The separation of the adaptation strategies from the source code of the eLearning software enables a system using the framework to quickly implement a variety of adaptation strategies. This work is a part of our more general ongoing work on the design of a framework for adaptive content delivery. parts of the framework discussed in this paper have been imulemented in a commercial eLearning engine.

• Proceedings of the Korean Information Science Society Conference
• /
• 2011.06a
• /
• pp.302-304
• /
• 2011

Rule based machine learning techniques are very important in our real world now. We can list out some important application which we can apply rule based machine learning algorithm such as medical data mining, business transaction mining. The different between rules based machine learning and model based machine learning is that model based machine learning out put some models, which often are very difficult to understand by expert or human. But rule based techniques output are the rule sets which is in IF THEN format. For example IF blood pressure=90 and kidney problem=yes then take this drug. By this way, medical doctor can easy modify and update some usable rule. This is the scenario in medical decision support system. Currently, Rough set is one of the most famous theory which can be used for produce the rule. LEM2 is the algorithm use this theory and can produce the small set of rule on the database. In this paper, we present an improvement of LEM2 algorithm which incorporates the variable precision techniques.

• Journal of Engineering Education Research
• /
• v.10 no.3
• /
• pp.109-124
• /
• 2007

With the advance of the Internet and computing technologies, e-learning is now a hot issue worldwide for providing the effective learning on the cyber-space. However, most of existing e-learning contents have been developed mainly based on text, including simple multimedia elements such as images, animations, and voices. This paper suggests a method effective for the computer programming e-learning. The method is based on program visualization using flowcharts. It features the stepwise hierarchical program visualization on the level of statements, the flowchart based visualization for control constructs of languages, visualization over whole programs, visualization compared with source codes, and interaction with users. Finally, we implement a system to realize the suggested method, and execute it for an example program.

• Journal of Institute of Control, Robotics and Systems
• /
• v.13 no.12
• /
• pp.1252-1256
• /
• 2007

U-learning (ubiquitous learning) systems, which deliver learning materials anytime and anywhere, allow learners to watch live lectures on PDAs, tablet PCs and notebook computers via broadband and wireless Internet. These systems have various problems; first, terminal devices are expensive, and it is difficult to maintain their efficiencies. Secondly, Internet does not guarantee quality of service (QoS), and in general it does not provide real-time services. Finally, the security of these systems is weaker in a local network than in an external network. The USB-based terminal system based on the OSGi service platform was designed as a ubiquitous system, in order to solve those problems. The USB terminals, used in this system, are inexpensive, and it is easy to maintain their performances. Also, this system solves the problems of security in a local network and provides guaranteed QoS. To accomplish this, the number of USB terminals connected to the system has to be limited according to the formula proposed in our paper. This system uses the OSGi specification as a middleware. It supports the discovery mechanism of the USB terminals, maintenance and administration of the system. Finally, this paper shows a driver's license testing system as an example u-learning application1.

• School Mathematics
• /
• v.7 no.3
• /
• pp.303-318
• /
• 2005

If students can use mathematics to solve their problems and learn the mathematical knowledge through it, they may think mathematics useful and valuable. This study is for the teaching through problem solving in mathematics education, which I consider in terms of the problem-based learning and mathematical modeling. 1 think mathematical modeling is applied to teaching mathematics as a problem-based learning. So I developed the teaching model, and showed the example that students learn the formal and hierarchic mathematics through mathematical modeling.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.7
• /
• pp.2359-2376
• /
• 2022

With the exponential growth of medical image big data represented by high-resolution CT images(CTI), the high-resolution CTI data is of great importance for clinical research and diagnosis. The paper takes lung CTI as an example to study. Retrieving answer CTIs similar to the input one from the large-scale lung CTI database can effectively assist physicians to diagnose. Compared with the conventional content-based image retrieval(CBIR) methods, the CBIR for lung CTIs demands higher retrieval accuracy in both the contour shape and the internal details of the organ. In traditional supervised deep learning networks, the learning of the network relies on the labeling of CTIs which is a very time-consuming task. To address this issue, the paper proposes a Weakly Supervised Similarity Evaluation Network (WSSENet) for efficiently support similarity analysis of lung CTIs. We conducted extensive experiments to verify the effectiveness of the WSSENet based on which the CBIR is performed.