• Journal of Information Processing Systems
• /
• 제17권3호
• /
• pp.441-452
• /
• 2021

Federated learning provides an efficient integrated model for distributed data, allowing the local training of different data. Meanwhile, the goal of multi-task learning is to simultaneously establish models for multiple related tasks, and to obtain the underlying main structure. However, traditional federated multi-task learning models not only have strict requirements for the data distribution, but also demand large amounts of calculation and have slow convergence, which hindered their promotion in many fields. In our work, we apply the rank constraint on weight vectors of the multi-task learning model to adaptively adjust the task's similarity learning, according to the distribution of federal node data. The proposed model has a general framework for solving optimal solutions, which can be used to deal with various data types. Experiments show that our model has achieved the best results in different dataset. Notably, our model can still obtain stable results in datasets with large distribution differences. In addition, compared with traditional federated multi-task learning models, our algorithm is able to converge on a local optimal solution within limited training iterations.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제18권1호
• /
• pp.30-45
• /
• 2024

This study introduces an effective method for predicting individual local tax delinquencies using prevalent machine learning and deep learning algorithms. The evaluation of credit risk holds great significance in the financial realm, impacting both companies and individuals. While credit risk prediction has been explored using statistical and machine learning techniques, their application to tax arrears prediction remains underexplored. We forecast individual local tax defaults in Republic of Korea using machine and deep learning algorithms, including convolutional neural networks (CNN), long short-term memory (LSTM), and sequence-to-sequence (seq2seq). Our model incorporates diverse credit and public information like loan history, delinquency records, credit card usage, and public taxation data, offering richer insights than prior studies. The results highlight the superior predictive accuracy of the CNN model. Anticipating local tax arrears more effectively could lead to efficient allocation of administrative resources. By leveraging advanced machine learning, this research offers a promising avenue for refining tax collection strategies and resource management.

• Journal of information and communication convergence engineering
• /
• 제22권1호
• /
• pp.23-32
• /
• 2024

This study aims to improve the quality of distance learning by developing webcast-based e-learning media and integrating it into an e-learning platform for functional job training purposes at the National Research and Innovation Agency, Indonesia. This study uses a Technology Acceptance Model (TAM) to assess and predict user perceptions of information systems using webcast platforms as an alternative to conventional applications. The research method was an online survey using Google Forms. Data collected from 136 respondents involved in practical job training were analyzed using structural equation modeling to test the technology acceptance model. The results showed that the proposed model effectively explained the variables associated with the adoption of web-based e-learning during the COVID-19 pandemic in Indonesia for participants engaged in functional job training. These findings suggest that users' perceptions of ease of use, usefulness, benefits, attitudes, intentions, and webcast usage significantly contribute to the acceptance and use of a more effective and efficient webcast-based e-learning platform.

• International Journal of Computer Science & Network Security
• /
• 제21권1호
• /
• pp.125-136
• /
• 2021

The development of systems that can generate automatically instructional material is a challenging goal for the e-learning community. These systems pave the way towards large scale e-learning deployment as they produce instruction on-demand for users requesting to learn about any topic, anywhere and anytime. However, realizing such systems is possible with the availability of vast repositories of web information in different formats that can be searched, reused and integrated into information-rich environments for interactive learning. This paradigm of learning relieves instructors from the tedious authoring task, making them focusing more on the design and quality of instruction. This paper presents a mobile learning system (Mole) that supports the generation of instructional material in M-Learning (Mobile Learning) contexts, by reusing and integrating heterogeneous hypermedia web resources. Mole uses open hypermedia repositories to build a Learning Web and to generate learning objects including various hypermedia resources that are adapted to the user context. Learning is delivered through a nice graphical user interface allowing the user to navigate conveniently while building their own learning path. A test case scenario illustrating Mole is presented along with a system evaluation which shows that in 90% of the cases Mole was able to generate learning objects that are related to the user query.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제6권3호
• /
• pp.869-885
• /
• 2012

The purpose of this study is to present innovative approaches for u-Learning environment in public education with Digital Textbook. The Korean Government has been making efforts to introduce the u-Learning environment to maximize the learning effect in public education with Digital Textbook. However, there are only a few studies that analyze the effectiveness of u-Learning environment and Digital Textbook. This paper reviews the current status of u-Learning environment in Korea and analyzes the satisfaction level with Digital Textbooks. The first survey regarding technological factors was collected from 197 students. The results of the survey revealed that the level of satisfaction has declined over a year. The weakness of the study is that the sample frame is insufficient and survey questions did not reflect diverse factors of learning effectiveness. To supplement these shortcomings, 2,226 students were asked about learning performance. The results of the survey showed that the satisfaction with Digital Textbooks is much higher than that of paper textbooks. However, this paper is limited to u-Learning environments in public education. Therefore, research needs to be improved by reflecting both public and private sectors of education in following studies. This paper suggests useful guidelines to educators in improving their u-Learning environment.

• 정보기술과데이타베이스저널
• /
• 제3권2호
• /
• pp.75-96
• /
• 1996

This study is intended to compare the effectiveness of the neural network inductive learning model with a vector space model in information retrieval. As a result, searches responding to incomplete queries in the neural network inductive learning model produced a higher precision and recall as compared with searches responding to complete queries in the vector space model. The results show that the hybrid methodology of integrating an inductive learning technique with the neural network model can help solve information retrieval problems that are the results of inconsistent indexing and incomplete queries--problems that have plagued information retrieval effectiveness.

• Journal of Information Science Theory and Practice
• /
• 제8권4호
• /
• pp.55-66
• /
• 2020

The purpose of this study is to examine the relationship between collaborative information seeking and users' learning style preferences and their experience of information systems. The study investigates the role of four different factors including learning style, task complexity, and user experience in collaborative information seeking in digital environments. Sixty participants (30 pairs) were randomly chosen from volunteer graduate students of Kharazmi University (Iran). Participants completed Kolb's learning style questionnaire and a user experience questionnaire and then performed two information seeking tasks (one simple and one difficult) in a lab setting. They could exchange information with their partners or a librarian using Skype. The sessions were recorded using Camtasia. The results showed that with an increase in task difficulty, collaborative information seeking activities increased and more interactions with partners and the librarian occurred. The number of executive help-seeking requests was higher than the number of instrumental help-seeking requests. This research confirms that learning style is related to the way users interact with the digital library and help seeking. The research showed that in difficult tasks, the differences among users with different learning styles become more evident, and that generally interactions increase in more difficult tasks. Among the learning styles, the accommodating style had the highest number of relationships with collaborative information seeking variables. Most of the statistically significant relationships between users' prior computer experience and collaborative information seeking variables were related to the time variable.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2005년도 ICCAS
• /
• pp.941-946
• /
• 2005

In this paper an active random noise control using adaptive learning rate neural networks is presented. The adaptive learning rate strategy increases the learning rate by a small constant if the current partial derivative of the objective function with respect to the weight and the exponential average of the previous derivatives have the same sign, otherwise the learning rate is decreased by a proportion of its value. The use of an adaptive learning rate attempts to keep the learning step size as large as possible without leading to oscillation. It is expected that a cost function minimize rapidly and training time is decreased. Numerical simulations and experiments of active random noise control with the transfer function of the error path will be performed, to validate the convergence properties of the adaptive learning rate Neural Networks. Control results show that adaptive learning rate Neural Networks control structure can outperform linear controllers and conventional neural network controller for the active random noise control.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 2005년도 한국컴퓨터종합학술대회 논문집 Vol.32 No.1 (A)
• /
• pp.79-81
• /
• 2005

최근 SCORM, Dublin Core등의 국제 표준 메타데이터와 함께, 세계 사실 표준이라 할 수 있는 IMS와 IEEE/LTSC의 LOM이 e-Learning의 특성을 반영한 메타데이터로서 현재 국$\cdot$내외적으로 많은 e-Learning 업체 및 기관에서 활용되고 있다(5). 그러나 LOM에서 정의한 메타데이터는 멀티플랫폼 환경을 고려하지 않고 있고, 제작 및 유통되고 있는 대부분의 e-Learning 콘텐트는 멀티미디어 특성에 대한 메타데이터 요소가 부족한 실정이다. 따라서 , 본 논문에서는 멀티플랫폼 환경에서 e-Learning학습을 지원하기 위해, 메타데이터 및 e-Learning 업체의 Requirement를 조사,분석하고 e-Learning 국제 표준 메타데이터와 플랫폼의 디바이스 특성을 반영하여, 기본적인 PC(Personal Computer) 환경을 포함한 모바일 기기 환경과 디지털TV 환경을 고려한 멀티플랫폼 e-Learning 메타데이터(Multi-platform e-Learning Metadata)를 제안하였다.

• 디지털산업정보학회논문지
• /
• 제13권2호
• /
• pp.1-11
• /
• 2017

In this paper, we design and implement PADIL(Prediction And Detection of Information Leakage) system that predicts and detect information leakage behavior of insider by analyzing network traffic and applying a variety of machine learning methods. we defined the five-level information leakage model(Reconnaissance, Scanning, Access and Escalation, Exfiltration, Obfuscation) by referring to the cyber kill-chain model. In order to perform the machine learning for detecting information leakage, PADIL system extracts various features by analyzing the network traffic and extracts the behavioral features by comparing it with the personal profile information and extracts information leakage level features. We tested various machine learning methods and as a result, the DecisionTree algorithm showed excellent performance in information leakage detection and we showed that performance can be further improved by fine feature selection.