지원사업
학술연구/단체지원/교육 등 연구자 활동을 지속하도록 DBpia가 지원하고 있어요.
커뮤니티
연구자들이 자신의 연구와 전문성을 널리 알리고, 새로운 협력의 기회를 만들 수 있는 네트워킹 공간이에요.
논문 기본 정보
- 자료유형
- 학술대회자료
- 저자정보
- 저널정보
- Korean Institute of Information Scientists and Engineers 한국정보과학회 학술발표논문집 한국정보과학회 2011가을 학술발표논문집 제38권 제2호(C)
- 발행연도
- 2011.11
- 수록면
- 13 - 16 (4page)
이용수
초록· 키워드
오류제보하기
We proposed a method for finding alternative clusterings of a dataset based on feature selection. We exploited the feature distance to find out the important features for data clusterings. We weighted the important features for the alternative clustering against those of the original one to make the alternative clustering is more likely to be found than the original clustering; but we still kept the other unimportant features so that data points stayed close to each other in terms of these features to reserve the quality. We tested our approach with six others on our synthetic dataset and UCI datasets: Segmentation, Vehicle, Vowel, Ionosphere and Glass. Our approach was the most stable one as it resulted in the best JI and DI for most of the tests. Our results showed that features are important for data clustering; this result is very useful in data mining based on data features.
목차
Abstract
1 Introduction
2 Previous works
3 Materials and Methods
4 Results
5 discussions
6 conclusion
7 Acknowledgments
References
참고문헌 (0)
참고문헌 신청이 논문의 저자 정보
최근 본 자료
댓글(0)
0