http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Multi-Task Support Vector Machine for Data Classification
Yunyan Song,Wenxin Zhu 보안공학연구지원센터 2016 International Journal of Signal Processing, Image Vol.9 No.7
Multi-task Learning (MTL) algorithms aim to improve the performance of several learning methods through shared information among all tasks. One particularly successful instance of multi-task learning is its adaptation to support vector machine (SVM). Recently advances in large-margin learning have shown that their solutions may be misled by the spread of data and preferentially separate classes along large spread directions. In this paper, we propose a novel formulation for multi-task learning by extending the recently published relative margin machine algorithm to the multi-task learning paradigm. The new method is an extension of support vector machine for single task learning. The objective of our algorithm is to obtain a different predictor for each task while taking into account the fact that the tasks are related as well as the spread of the data. We test the proposed method experimentally using real data. The experiments show that the proposed method performs better than existing multi-task leaning with SVM and single-task leaning with SVM.
Least Squares Fuzzy One-class Support Vector Machine for Imbalanced Data
Jingjing Zhang,Kuaini Wang,Wenxin Zhu,Ping Zhong 보안공학연구지원센터 2015 International Journal of Signal Processing, Image Vol.8 No.8
Based on fuzzy one-class support vector machine (SVM) and least squares (LS) one-class SVM, we propose an LS fuzzy one-class SVM to deal with the class imbalanced problem. The LS fuzzy one-class SVM applies a fuzzy membership to each sample and attempts to solve the modified primal problem. Hence, we just need to solve a system of linear equations as opposed solving the quadratic programming problem (QPP) in fuzzy one-class SVM, which leads to an extremely simple and fast algorithm. Numerical experiments on several benchmark data sets demonstrate the feasibility and effectiveness of the proposed algorithm.