• 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 KIISE:Software and Applications
• /
• v.31 no.9
• /
• pp.1204-1217
• /
• 2004

Numerous statistical and machine learning techniques have been studied for automatic text classification. However, because they train the classifiers using only feature vectors of documents, ambiguity between two possible categories significantly degrades precision of classification. To remedy the drawback, we propose a new method which incorporates relationship information of categories into extant classifiers. In this paper, we first perform the document classification using the k-NN classifier which is generally known for relatively good performance in spite of its simplicity. We employ the relationship information from an object-based thesaurus to reduce the ambiguity. By referencing various relationships in the thesaurus corresponding to the structured categories, the precision of k-NN classification is drastically improved, removing the ambiguity. Experiment result shows that this method achieves the precision up to 13.86% over the k-NN classification, preserving its recall.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.11a
• /
• pp.864-867
• /
• 2021

뉴스, SNS 등의 인터넷 댓글은 익명으로 의견을 자유롭게 개진할 수 있는 반면 댓글의 익명성을 악용하여 비방이나 험담을 하는 악성 댓글이 여러 분야에서 사회적 문제가 되고 있다. 해당 문제를 해결하기 위해 AI를 활용한 댓글 분류 알고리즘을 개발하려는 많은 노력들이 이루어지고 있지만, 댓글 분류 모델에 사용되는 AI는 오버피팅의 문제로 인해 댓글 분류에 대한 정확도가 떨어지는 문제점을 가지고 있다. 이에 본 연구에서는 크라우드소싱을 활용하여 오버피팅으로 인한 악성 댓글 분류 및 판단 정확도 저하 문제를 개선한 크라우드소싱 기반 딥러닝 분류 알고리즘(Deep Learning Classification Algorithm Based on Crowdsourcing: DCAC)과 해당 알고리즘을 사용한 시스템을 제안한다. 또한, 실험을 통해 오버피팅으로 낮아진 판단 정확도를 증가시키는 데 제안된 방법이 도움이 되는 것을 확인하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2008.05a
• /
• pp.968-970
• /
• 2008

기존에 트래픽 분류 방법으로 payload 분석이나 well-known port를 이용한 방법을 많이 사용했다. 하지만 동적으로 변하는 애플리케이션이 늘어남에 따라 기존 방법으로 애플리케이션 트래픽 분류가 어렵다. 이러한 문제의 대안으로 Machine Learning(ML) 알고리즘을 이용한 애플리케이션 트래픽 분류방법이 연구되고 있다. 기존의 논문에서는 일정 시간동안 수집한 data set을 사용하기 때문에 적게 발생한 애플리케이션은 제대로 분류하지 못하여도 전체적으로는 좋은 성능을 보일 수 있다. 본 논문에서는 이러한 문제를 해결하기 위해 각 애플리케이션마다 동일한 수의 data set을 수집하여 애플리케이션 트래픽을 분류하는 방법을 제시한다. ML 알고리즘 중 J48, REPTree, BayesNet, NaiveBayes, Multilayer Perceptron 알고리즘을 이용하여 애플리케이션 트래픽 분류의 정확도를 비교한다.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.4 no.4
• /
• pp.1-7
• /
• 2001

Remotely sensed data is the most fundamental data in acquiring the GIS informations, and may be analyzed to extract useful thematic information. Multi-spectral classification is one of the most often used methods of information extraction. The actual multi-spectral classification may be performed using either supervised or unsupervised approaches. This paper analyze the effect of assigning clever initial values to image classes on the performance of parallelepiped classification algorithm, which is one of the supervised classification algorithms. First, we investigate the effect on serial computing model, then expand it on MIMD(Multiple Instruction Multiple Data) parallel computing model. On serial computing model, the performance of the parallel pipe algorithm improved 2.4 times at most and, on MIMD parallel computing model the performance improved about 2.5 times as clever initial values are assigned to image class. Through computer simulation we find that initial values of image class greatly affect the performance of parallelepiped classification algorithms, and it can be improved greatly when classes on both serial computing model and MIMD parallel computation model.

• The Journal of the Acoustical Society of Korea
• /
• v.26 no.8
• /
• pp.390-396
• /
• 2007

In this letter, we propose a novel approach to improve the performance of speech/music classification for the selectable mode vocoder(SMV) of 3GPP2 using the Gaussian mixture model(GMM) which is based on the expectation-maximization(EM) algorithm. We first present an effective analysis of the features and the classification method adopted in the conventional SMV. And then feature vectors which are applied to the GMM are selected from relevant Parameters of the SMV for the efficient speech/music classification. The performance of the proposed algorithm is evaluated under various conditions and yields better results compared with the conventional scheme of the SMV.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.32 no.10C
• /
• pp.965-974
• /
• 2007

In this paper, we proposed the feature selection algorithm for multi-class genre classification. In our proposed algorithm, we developed GMM separation score based on Gaussian mixture model for measuring separability between two genres. Additionally, we improved feature subset selection algorithm based on sequential forward selection for multi-class genre classification. Instead of setting criterion as entire genre separability measures, we set criterion as worst genre separability measure for each sequential selection step. In order to assess the performance proposed algorithm, we extracted various features which represent characteristics such as timbre, rhythm, pitch and so on. Then, we investigate classification performance by GMM classifier and k-NN classifier for selected features using conventional algorithm and proposed algorithm. Proposed algorithm showed improved performance in classification accuracy up to 10 percent for classification experiments of low dimension feature vector especially.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2019.05a
• /
• pp.367-370
• /
• 2019

본 논문은 구매이력 데이터에서 상품간의 분류 체계를 고려하여 시퀀스 간의 유사도를 계산하는 새로운 방법을 제안한다. 시퀀스란 두 항목간의 순서가 존재하는 데이터를 의미한다. 항목 간의 선후관계가 중요한 시퀀스 데이터에서는 두 시퀀스 간의 유사도를 정확히 정의하는 것이 중요하다. 본 논문에서는 대표적인 시퀀스 유사도 측정 알고리즘인 편집 거리 알고리즘을 활용하여 구매이력 데이터에서 시퀀스 간의 유사도를 정의한다. 상품은 상품의 특성에 따라 항목 분류 체계에서 여러 범주로 분류된다. 이 경우 기존의 편집 거리 알고리즘에서 문자의 일치유무에 따라 단순히 0 또는 1을 부여하는 것은 부정확하다. 따라서 본 논문은 편집 거리 알고리즘의 수정 연산 중 대체 연산 비용 계산 시 항목 분류 트리를 사용하여 연산 비용이 0 에서 1 사이의 값을 가지도록 세분화하였다. 실험 결과 제안 방법은 대체 연산 비용 계산 시 두 문자가 다르면 단순히 1 을 부여하는 기존의 편집 거리 알고리즘에 비해 시퀀스 간의 유사도를 더 정확하게 계산함을 확인하였다.

• The KIPS Transactions:PartB
• /
• v.9B no.2
• /
• pp.202-214
• /
• 2002

In this paper, we propose a modified FCM (MFCM) algorithm to solve the problems of the FCM algorithm and the fuzzy clustering algorithm using an average intracluster distance (FCAID). The MFCM algorithm grants the regular grade of membership in the small size of cluster. And it clears up the convergence problem of objective function because its objective function is designed according to the grade of membership of it, verified, and used for clustering data. So, it can solve the problem of the FCM algorithm in different size of cluster and the FCAID algorithm in the convergence problem of objective function. To verify the MFCM algorithm, we compared with the result of the FCM and the FCAID algorithm in data clustering. From the experimental results, the MFCM algorithm has a good performance compared with others by classification entropy.

• The KIPS Transactions:PartB
• /
• v.16B no.3
• /
• pp.233-238
• /
• 2009

This paper presents an approach to the classifications of small size document using the instance-based feature filtering Relief-F algorithm. In the document classifications, we have not always good classification performances of small size document included a few features. Because total number of feature in the document set is large, but feature count of each document is very small relatively, so the similarities between documents are very low when we use general assessment of similarity and classifiers. Specially, in the cases of the classification of web document in the directory service and the classification of the sectors that cannot connect with the original file after recovery hard-disk, we have not good classification performances. Thus, we propose the Extended Relief-F(ERelief-F) algorithm using instance-based feature filtering algorithm Relief-F to solve problems of Relief-F as preprocess of classification. For the performance comparison, we tested information gain, odds ratio and Relief-F for feature filtering and getting those feature values, and used kNN and SVM classifiers. In the experimental results, the Extended Relief-F(ERelief-F) algorithm, compared with the others, performed best for all of the datasets and reduced many irrelevant features from document sets.