Research on classifying technique for imbalanced dataset based on Support Vector Machines

Zhi-ming Yang; Yu Peng; Xi-yuan Peng

doi:10.1117/12.807706

12 January 2009 Research on classifying technique for imbalanced dataset based on Support Vector Machines

Zhi-ming Yang, Yu Peng, Xi-yuan Peng

Proceedings Volume 7133, Fifth International Symposium on Instrumentation Science and Technology; 713320 (2009) https://doi.org/10.1117/12.807706
Event: International Symposium on Instrumentation Science and Technology, 2008, Shenyang, China

Abstract

It is shown that SVM can be ineffective in classifying the minority samples, when it is applied to the problem of learning from imbalanced datasets. To remedy this problem, this paper analyzes the true reason of negative effect to SVM classifier caused by data imbalance firstly. Based on this, a new method of shifting classifying hyperplane in the feature space is proposed, and its implementation method-Boundary Movement based on Sample Cutting Technique (BMSCT) is also described. Through theoretical analysis and empirical study, we show that our method augments the classification accuracy rate effectively without increasing the computation complexity.

Citation Download Citation

Zhi-ming Yang, Yu Peng, and Xi-yuan Peng "Research on classifying technique for imbalanced dataset based on Support Vector Machines", Proc. SPIE 7133, Fifth International Symposium on Instrumentation Science and Technology, 713320 (12 January 2009); https://doi.org/10.1117/12.807706

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available