• Journal of the Korean Institute of Intelligent Systems
• /
• v.16 no.6
• /
• pp.736-741
• /
• 2006

The design of a classification system generally involves data acquisition module, learning module and decision module, considering their functions and it is often an important component of intelligent systems. The learning module provides a priori information and it has been playing a key role for the classification. The conventional learning techniques for classification are based on a winner take all fashion which does not reflect the description of real data where boundarues might be fuzzy Moreover they need all data for the learning of its problem domain. Generally, in many practical applications, it is not possible to prepare them at a time. In this paper, we design an adaptive classification model using incremental training fuzzy neural networks, FNN-I. To have a more useful information, it introduces the representation and membership degree by fuzzy theory. And it provides an incremental learning algorithm for continuously gathered data. We present tie experimental results on computer virus data. They show that the proposed system can learn incrementally and classify new viruses effectively.

• Proceedings of the Korean Information Science Society Conference
• /
• 2002.04b
• /
• pp.301-303
• /
• 2002

컴퓨터가 널리 보급되고 인터넷이 발전함에 따라 수없이 많은 정보가 디지털 형태로 생산되고 있다. 이러한 정보를 사람이 일일이 가공하고 분류하기에는 한계가 있으므로 자동으로 문서를 분류하고자 하는 연구가 대두되었다. 문서를 자동으로 분류하기 위해 기계학습 방법이 많이 이용되고 있다. 기계학습방법을 이용한 문서분류가 좋은 성능을 내기 위해서는 충분한 양의 학습데이터가 필요하다. 학습데이터를 만들기 위해서는 사람이 일일이 분류해야 하므로, 비용이 많이 든다. 본 논문에서는 적은양의 labeled 데이터로부터 시작하여, 점증적으로 unlabeled 데이터를 학습에 참여시킴으로써, 문서분류의 성능을 높이고자 한다. 실험을 통해 Unlabeled 문서데이터를 사용한 것이 좋은 성능을 보였음을 알 수 있다.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.13 no.3
• /
• pp.266-274
• /
• 2003

This paper considers an incremental support vector learning for the abnormality detection problems. One of the most well-known support vector learning methods for abnormality detection is the so-called SVDD(support vector data description), which seeks the strategy of utilizing balls defined on the kernel feature space in order to distinguish a set of normal data from all other possible abnormal objects. The major concern of this paper is to modify the SVDD into the direction of utilizing the relation between the optimal solution and incrementally given training data. After a thorough review about the original SVDD method, this paper establishes an incremental method for finding the optimal solution based on certain observations on the Lagrange dual problems. The applicability of the presented incremental method is illustrated via a design example.

• Journal of Internet Computing and Services
• /
• v.4 no.5
• /
• pp.69-76
• /
• 2003

In this paper we show a new learning algorithm for pattern classification. This algorithm considered a scheme to find a solution to a problem of incremental learning algorithm when the structure becomes too complex by noise patterns included in learning data set. Our approach for this problem uses a pruning method which terminates the learning process with a predefined criterion. In this process, an iterative model with 3 layer feedforward structure is derived from the incremental model by an appropriate manipulations. Notice that this network structure is not full-connected between upper and lower layers. To verify the effectiveness of pruning method, this network is retrained by EBP. From this results, we can find out that the proposed algorithm is effective, as an aspect of a system performence and the node number included in network structure.

• Proceedings of the Korean Information Science Society Conference
• /
• 2001.04b
• /
• pp.334-336
• /
• 2001

세선화 알고리즘의 간접 기법으로 제시된 자기구성 특징 그래프(Self-Organizing feature Graph : SOG) 기법은 안정된 세선화 결과를 가지는 장점이 있으나 학습 알고리즘에서 전체 노드를 재정돈하는 과정이 내포되어 있다. 본 본문에서는 학습 알고리즘의 재정돈 과정을 대신하는 점증적 정돈기법을 제안하고 이 기법을 세선화 알고리즘에 결합하여 실험하고 분석하였다. 제안된 알고리즘은 기존의 SO를 적용한 결과와 같은 우수한 세선화 결과를 얻으며 학습시간은 O((logM)$^3$)인 복잡도를 가진다.

• Proceedings of the Korean Information Science Society Conference
• /
• 1999.10b
• /
• pp.6-8
• /
• 1999

시뮬레이션 환경이나 실제 환경에서 이동 로봇 제어기를 진화 알고리즘으로 만들어내는 연구가 최근 활발하다. 이전의 연구에서는 기존의 단순한 진화 알고리즘이 환경에 제한된 제어기를 만들어 내는 문제점을 해결하기 위한 방법으로 셀룰라 오토마타 기반 신경망의 점증적 진화방법을 제시하였다. 점증적 진화 방법은 초기에 간단한 행동으로 해결할 수 있는 환경에 맞도록 제어기를 진화시킨 다음, 점차 복잡한 행동이 요구되는 환경에서 제어기를 점증적으로 진화시킨다. 실험결과, 점증적 진화의 방법이 좀 더 효율적으로 로봇을 진화시키고 환경의 변화에 보다 강한 것을 알 수 있었다. 그러나 이전연구에서의 점증적 진화 방법은 한 단계에서 진화가 끝난 후 다음 단계로 넘어갈 개체를 사람이 선택해야 하는 문제가 있었다. 본 논문에서는 이러한 문제점을 해결하기 위한 다양한 방법을 제시하고 실험을 통해 그 유용성을 보이고자 한다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.9 no.1
• /
• pp.40-44
• /
• 2005

Applying artificial intelligence, machine learning and data mining techniques to intrusion detection system are increasing. But most of researches are focused on improving the performance of classifier. These classifiers are performed by batch way and it is not proper method for realtime intrusion detection system. We propose an incremental feature extraction and classification technique for realtime intrusion detection system. Applying proposed system to KDD CUP 99 data, experimental result shows that it has similar capability compared to batch way intrusion detection system.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2001.04b
• /
• pp.1029-1032
• /
• 2001

자기구성 특징지도(Self-Organizing feature Map : SOM) 및 $L^*$ 등의 자가 학습 신경망의 알고리즘들은 학습 결과 중에 바람직하지 못한 뒤틀림 현상(cross-over)을 생성하게 되므로 재학습으로 인한 전반적인 학습 시간의 지연을 초래한다. 이 논문에서는 비교적 학습 속도가 빠른 $L^*$의 점증적 학습 구조를 기본으로 하여 뒤틀림 현상 방지를 목적으로 초기 학습 단계에서 학습 가중치들의 노드들을 재조정하는 개선된 알고리즘을 제안한다. 이러한 알고리즘의 실험 결과는 모두 정상적인 학습 결과를 보이고 학습의 시행 착오적인 재실행이 없으므로 전반적인 학습 속도는 기존의 알고리즘보다 빠르게 됨을 보인다.

• Journal of the Korea Computer Industry Society
• /
• v.5 no.9
• /
• pp.903-910
• /
• 2004

From the learner competency-based education, the necessity of the learner-oriented education and the mutual cooperation among the leamers has been increased. An internship-based education system will achieve a progressive development in understanding and the competency for the job to be performed by the mutual interactions between the leamers and between the learners and the lecturer. This paper presents the project-based learning system in order to facilitate and achieve the goals of the industrial business processing system implementations in the ongoing popular Internet environments. In the proposed internship-based system, the lecturer presents a problem to be solved and leads the education as a supervisor.

• The KIPS Transactions:PartA
• /
• v.9A no.4
• /
• pp.621-630
• /
• 2002

Web based lecture & learning system has changed the conventional learner-oriented educational environment by adding mutual interactions. From the learner competency-based education. the necessity of the learner-oriented education and the mutual cooperation among the learners has been increased. A project-based education system will achieve a progressive development in understanding and the competency for the job to be performed by the mutual interactions between the learners and between the learners and the lecturer. This paper presents the project-based learning system in order to facilitate and achieve the goals of the industrial business processing system implementations in the ongoing popular Internet environments. In the proposed project-based system, the lecturer presents a problem to be solved and leads the education as a supervisor The learner can maximize the efficiency for study by taking advantage of the learner-oriented comparison study and the patterning.