期刊名称：International Journal of Advanced Research in Computer Engineering & Technology (IJARCET)
印刷版ISSN：2278-1323
出版年度：2015
卷号：4
期号：9
页码：3734-3740
出版社：Shri Pannalal Research Institute of Technolgy
摘要：Multi-Instance Heterogeneous Classifiers with Extended Space forest (MIHC_ES)is a new method for Feature Set Generation along with efficient heterogeneous ensembling of classifier employed for linear classification problem. Extended Space feature generation is new and efficient system of generating new feature from original feature set. Ensemble classification system consists of multiple classifiers in which each classifier set consist of classifier instance of same type. In heterogeneous ensembling each classifier in the classifier set have multiple instance of same type of classifier together with different heterogeneous classifiers used for active learning. This set of Heterogeneous classifier within ensemble is capable of changing number of instances of each classifier type within the ensemble based maximum and minimum accuracy achieved . The three major algorithm adopted for this experiment is collaborating extended space forest and stably sized heterogeneous ensembling of classifier and rotation forest .For Heterogeneous Ensembles (HE), experimental evaluations show that HE constructs heterogeneous ensembles that outperform homogeneous ensembles composed of any one of the classifier types, as well as it outperforms AHE on many analysis data set. We in this system leveraged the advantage of AHE over other methods by adapting instances of classifier type in overall in the ensemble during learning and the target data set is composed of target class labels.
