• Journal of Intelligence and Information Systems
• /
• v.4 no.2
• /
• pp.1-21
• /
• 1998

귀납적 학습방법들의 분류성능을 비교 평가하기 위하여 대표적 분류문제의 하나인 신용평가 문제를 사용하였다. 분류기로서 사용된 귀납적 학습방법론들은 통계학의 다변량 판별분석(MDA), 기계학습 분야의 C4.5, 신경망의 다계층 퍼셉트론(MLP) 및 Cascade Correlation Network(CCN)의 4 가지이며, 학습자료로는 국내 3개 신용평가기관이 발표한 신용등급 및 공포된 재무제표를 사용하였다. 신용등급 예측의 정확도에 의한 분류성능을 평가하였는데 연도별 평가와 시계열 평가의 두 가지를 실시하였다. Cascade Correlation Network이 가장 좋은 분류성능을 보였지만 4가지 분류기들 사이에 통계적으로 유의한 차이는 발견되지 않았다. 이는 사용된 학습자료가 갖는 한계로 인한 것으로 추정되지만, 성능평가 과정에 있어 학습자료의 전처리 과정이 분류성과의 제고에 매우 유효함이 입증되었다.

• Journal of Intelligence and Information Systems
• /
• v.7 no.2
• /
• pp.51-63
• /
• 2001

In supervised machine learning, an induction algorithm, which is able to extract rules from data with learning capability, provides a useful tool for data mining. Practical induction algorithms are known to degrade in prediction accuracy and generate complex rules unnecessarily when trained on data containing superfluous features. Thus it needs feature subset selection for better performance of them. In feature subset selection on the induction algorithm, wrapper method is repeatedly run it on the dataset using various feature subsets. But it is impractical to search the whole space exhaustively unless the features are small. This study proposes a heuristic method that uses sensitivity analysis of neural networks to the wrapper method for generating rules with higher possible accuracy. First it gives priority to all features using sensitivity analysis of neural networks. And it uses the wrapper method that searches the ordered feature space. In experiments to three datasets, we show that the suggested method is capable of selecting a feature subset that improves the performance of the induction algorithm within certain iteration.

• Journal of The Korean Association of Information Education
• /
• v.22 no.5
• /
• pp.575-583
• /
• 2018

This study approached EPL learning with deductive teaching and learning methods and inductive teaching and learning methods which are grammar teaching and learning methods. In the entry site, lectures provided for elementary school students in grades 5 to 6 were set as deductive learning courses. Based on this, inductive learning process was developed and each learning process was composed of 12 periods. After conducting the research, EPL utilization evaluation, learning satisfaction and immersion test were conducted between the two groups. It was difficult to obtain statistically meaningful results between the two groups. However, in the three tests, the mean value of groups using inductive teaching and learning methods was high. If we construct a long-term learning process and conduct research, we think that statistically meaningful results are produced between the two groups.

• The KIPS Transactions:PartB
• /
• v.11B no.1
• /
• pp.83-90
• /
• 2004

Strategic games are missing special qualities of genre these days. Game engines neither reason about behaviors of computer objects nor have learning ability that can prepare countermeasure in variously command user's strategy. This paper suggests a strategic game engine that applies non-monotonic reasoning and inductive machine learning. The engine emphasizes three components -“user behavior monitor”to abstract user's objects behavior,“learning engine”to learn user's strategy,“behavior display handler”to reflect abstracted behavior of computer objects on game. Especially, this paper proposes two layered-structure to apply non-monotonic reasoning and inductive learning to make behaviors of computer objects that learns strategy behaviors of user objects exactly, and corresponds in user's objects. The engine decides actions and strategies of computer objects with created information through inductive learning. Main contribution of this paper is that computer objects command excellent strategies and reveal differentiation with behavior of existing computer objects to apply non-monotonic reasoning and inductive machine learning.

• Journal of Korea Multimedia Society
• /
• v.4 no.3
• /
• pp.280-285
• /
• 2001

Since KBANN (knowledge-based artificial neural network) combing the inductive learning algorithm and the analytical learning algorithm was proposed, several methods such as TopGen, TR-KBANN, THRE-KBANN which modify KBANN have been proposed. But these methods can be applied when there is a domain theory. The algorithm representing the problem into KBANN based on only the instances without domain theory is proposed in this paper. Domain theory represented into KBANN can be refined by THRE-KBANN. The performance of this algorithm is more efficient than the C4.5 in the experiment for some problem domains of inductive learning.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.10 no.3
• /
• pp.614-621
• /
• 2006

We have implemented a multiclassifier learning approach in a GA-based inductive learning environment that learns classification rules that are similar to rules used in PROSPECTOR. In the multiclassifier learning approach, a classification system is constructed with several classifiers that are obtained by running a GA-based learning system several times to improve the overall performance of a classification system. To implement the multiclassifier learning approach, we need a decision-making scheme that can draw a decision using multiple classifiers. In this paper, we introduce two decision-making schemes: one is based on combining posterior odds given by classifiers to each class and the other one is a voting scheme based on ranking assigned to each class by classifiers. We also present empirical results that evaluate the effect of the multiclassifier learning approach on the GA-based inductive teaming environment.

• Journal of Digital Convergence
• /
• v.19 no.9
• /
• pp.307-316
• /
• 2021

This study aims to explore the effects of deductive FFI and inductive FFI in L2 learners' grammatical achievement and their reaction to the grammar instruction. 84 students were placed into three groups: 29 given deductive FFI(DG), 28 receiving inductive FFI(IG), and 27 with traditional instruction(CG). All students completed pre/post tests and questionnaires, and took a delayed post test 9 weeks after the treatment. For statistical anlayses of all the quantitative data, a one-way ANOVA, paired samples T-test, and repeated measures ANOVA were performed. The results indicated that both deductive and inductive FFI affected learners' grammatical achievement and their achievement was sustained over time. Deductive FFI was more effective than inductive FFI, whereas the IG students more positively changed their attitudes and perceptions to the grammar instruction. These findings of the study imply that FFI should be valued in an Korean EFL classroom, which would contribute to further longitudinal research for its sustainability.

• Proceedings of the Korea Database Society Conference
• /
• 1999.06a
• /
• pp.291-300
• /
• 1999

본 논문은 인공신경망과 귀납적 학습방법 등의 인공지능 방법과 선물가격결정에 대한 기존 재무이론을 사용하여 일상어취로 표현되는 파생상품 가격예측 모형을 개발하는데 있다. 모형의 개발은 1단계로 인공신경망이나 기존의 선물가격결정이론(평균보 유비용모형이나 일반균형모형)을 이용하여 선물 가격을 예측한 후, 서로 비교 분석하여 인공신경망 모형의 우수성을 확인하였다. 귀납적 학습방법중 CART 알고리듬을 사용하여 If-Then 규칙을 생성하였다. 특히 실용적 측면에서 선물가격의 일상어휘화를 통한 모형개발을 여러 가지 방법으로 시도하였다. 이러한 선물가격 예측모형의 유용성은 일단 If-Then 규칙으로 표현되어 전문가의 판단에 확실한 이론적인 근거를 제시할 수 있는 장점이 있으며, 특히 의사결정지원시스템으로 활용화 될 경우 매우 유용한 근거자료로 활용될 수 있다. 이러한 선물가격 예측모형의 정확성은 분석표본과 검증표본으로 나누어 검증표본에서 세가지 기본모형(평균보유 비용모형, 일반균형모형, 인공신경망 모형)과 각 모형의 귀납적 학습방법 모형의 다른 3가지 어휘표현방법 3가지를 모형별로 비교 분석하였다. 분석결과 인공신경망모형은 상당한 예측력을 갖고 있는 것으로 판명되었으며, 특히 CART를 기반으로 한 일상어취 기반의 선물가격예측 모형은 예측력이 높은 것으로 나타났다.

• Proceedings of the Korean Information Science Society Conference
• /
• 1999.10b
• /
• pp.54-56
• /
• 1999

웹 에이전트는 사용자가 웹을 브라우징하는 행위를 모니터하여 사용자의 관심정보를 학습하고 사용자가 필요로 한느 웹 상의 정보를 제공하는 시스템이다. 웹 에이전트는 사용자의 관심정보를 추출하기 위해서 귀납적 기계학습을 수행한다. 이때, 학습의 효율을 높이기 위해서는 관련이 있는 문서들을 그룹화하여 학습 시스템에 제공하여야 한다. 본 논문에서는 비감독 개념 학습 알고리즘인 COBWEB을 이용하여 사용자가 관심을 표시한 문서들의 분류트리를 생성한다. 분류트리는 귀납적 기계학습 시스템의 입력으로 사용될 수 있는 형태가 아니므로 분류 트리의 분석과 문서 분류 후처리 작업을 통해서 문서 집합을 생성해야 한다. 이를 위해서는 분류트리를 분석하여 초기 클러스터를 생성하고, 유사한 클러스터들의 병합을 수행한다. 본 논문에서 제안하는 문서 자동 분류 방식은 비감독 개념 학습 알고리즘이 생성한 문서 분류 트리의 분석을 통해서 충분한 유사도와 적절한 수의 문서를 포함하는 초기 클러스터를 생성할 수 있다. 그러므로 문서 분류의 후처리 작업인 클러스터의 병합 작업에서 불필요한 작업을 제거함으로서 보다 효과적이고 합리적인 문서 분류 작업을 수행한다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.11 no.3
• /
• pp.619-626
• /
• 2007

Constructive induction is a technique to draw useful attributes from given primitive attributes to classify given examples more efficiently. Useful attributes are obtained from given primitive attributes by applying appropriate operators to them. The paper proposes a constructive induction approach for a GA-based inductive learning environment that learns classification rules that ate similar to rules used in PROSPECTOR from given examples. The paper explains our constructive induction approach in details, centering on operators to combine primitive attributes and methods to evaluate the usefulness of derived attributes, and presents the results of various experiments performed to evaluate the effect of our constructive induction approach on the GA-based learning environment.