For Full-Text PDF, please login, if you are a member of IEICE,|
or go to Pay Per View on menu list, if you are a nonmember of IEICE.
Specific Random Trees for Random Forest
Zhi LIU Zhaocai SUN Hongjun WANG
IEICE TRANSACTIONS on Information and Systems
Publication Date: 2013/03/01
Online ISSN: 1745-1361
Print ISSN: 0916-8532
Type of Manuscript: LETTER
Category: Artificial Intelligence, Data Mining
random forest, multiclass classification, specific random trees,
Full Text: PDF>>
In this study, a novel forest method based on specific random trees (SRT) was proposed for a multiclass classification problem. The proposed SRT was built on one specific class, which decides whether a sample belongs to a certain class. The forest can make a final decision on classification by ensembling all the specific trees. Compared with the original random forest, our method has higher strength, but lower correlation and upper error bound. The experimental results based on 10 different public datasets demonstrated the efficiency of the proposed method.