약한 분류기에 따른 Adaboosting알고리즘 성능 비교

• 영문명: Performance Evaluation of Adaboosting Algorithm according to Weak Classifier
• 발행기관: 한국IT마케팅학회
• 저자명: 권태윤(Tae Yoon Kwon) 권영만(Young Man Kwon)
• 간행물 정보: 『한국IT마케팅학회 논문집』2015 IT마케팅학회 논문집 Vol.1 No.1, 152~154쪽, 전체 3쪽
• 주제분류: 경제경영 > 경영학
• 파일형태: PDF
• 발행일자: 2015.11.20

영문 초록

There are many algorithms about Adaboost classification. Moreover, Data has many types of patterns and distribution. So it is hard to select the right algorithm to apply to real problem. According to these problems we need the best algorithm with less error rate to classification. In this paper we did experiment to find the best algorithm relying on data patterns. First, we made four data patterns as artificially. After then, our proposed algorithm used adaptive boosting(AdaBoost) with weak classifier as Perceptron and Decision Tree. As a result,Perceptron in Adaboost algorithm generally show less error rate in the all kind of data patterns.

목차

Abstract

I. 서론

II. 관련 연구

III. 실험 방법 및 결과

IV. 결론

참고문헌

키워드

퍼셉트론 의사결정나무 아다부스트 Perceptron DecisionTree Adaboost