作者: Kenji Watanabe , Takio Kurita
DOI: 10.1109/ICPR.2008.4761279
关键词:
摘要: In this paper, we propose a novel algorithm of multi-nominal logistic regression in which the locality regularization term is introduced. The defined by neighborhood information data set and preserved mapped feature space. By using standard benchmark datasets, it was shown that proposed gave higher recognition rates than linear SVM binary classification problems. for multi-class problem were also better general regression.