南京大学学报（自然科学版）2018，Vol.54Issue(1)：116-123,8.DOI:10.13232/j.cnki.jnju.2018.01.013
HSEC:基于聚类的启发式选择性集成
HSEC:Clustering based heuristic selelctive ensemble learning algorithm
摘要
Abstract
This paper introduces a clustering-based heuristic selective ensemble algorithm,called clustering based heuristic selective ensemble learning algrothm (HSEC).Ensemble learning obtains a better predictive through combining multiple weak models.However,with the increasing amount of the weak model,the complexity and training time become more complex.We use a clustering selective method to eliminate similarity classifiers to decrease the amount of weak models.Then,we select the best sequence set of classifiers based on the heuristic selective ensemble algorithm.We use multiprocessors to train model to solve the inefficiency of the selection classifier in the previous study for integration of learning,which could improve efficiency greatly.The experimental results show that the HSEC algorithm has made a certain boost in some indicators comparing with traditional classi-fication algorithm.关键词
集成学习/选择性集成学习/聚类/降维Key words
ensemble learning/selective ensemble learning/clustering/dimension reduction分类
信息技术与安全科学引用本文复制引用
郑丽容,洪志令..HSEC:基于聚类的启发式选择性集成[J].南京大学学报（自然科学版）,2018,54(1):116-123,8.基金项目
国家自然科学基金(31200769) （31200769）
