CAJ | 학술논문

支持向量机（support vector machine，SVM）是分类算法中集高效性、准确率和实时性于一体的分类方案。但由于在SVM分类决策的过程中，无关的分类器也参与了投票，使得方案的实时性和分类可靠性有一定程度的降低。提出了基于相似度的高效SVM网络流量识别方案（efficient SVM based on similarity，ESVMS）。ESVMS通过估算待分类实例可能所属的类别范围，排除SVM中那些无关分类器的投票决策。实验结果表明ESVMS较SVM分类准确度几乎没有降低，但分类实时性进一步提高。
지지향량궤（support vector machine，SVM）시분류산법중집고효성、준학솔화실시성우일체적분류방안。단유우재SVM분류결책적과정중，무관적분류기야삼여료투표，사득방안적실시성화분류가고성유일정정도적강저。제출료기우상사도적고효SVM망락류량식별방안（efficient SVM based on similarity，ESVMS）。ESVMS통과고산대분류실례가능소속적유별범위，배제SVM중나사무관분류기적투표결책。실험결과표명ESVMS교SVM분류준학도궤호몰유강저，단분류실시성진일보제고。
Support Vector Machine is a classification algorithm that combines high efficiency,high accuracy and real time.Theres a problem when SVM makes its decision for an un-labeled instance because uninvolved classifierspartici-pate in that affects SVMs real time performance and reliability.Thus,a method utilized Efficient SVM based on Simi-larity (ESVMS)for traffic classification was proposed.ESVMS estimates the classes that an un-labeled instances may belongs to as to kick out the uninvolved classifiers.Experimental results show that ESVMS holds the accuracy of SVMs and improves its real time performance.