CAJ | 학술논문

为了进一步提高SVM集成的泛化能力,提出了基于Choquet模糊积分的SVMs集成方法,综合考虑各个子SVM输出重要性,避免了现有SVM集成方法中忽略次要信息的问题。应用该方法,以高校的区域经济贡献度为例进行仿真试验,结果表明基于Choquet模糊积分的SVMs集成方法较基于Sugeno模糊积分SVMs集成方法和基于投票策略的SVMs集成方法具有更高的准确性。该方法是可行、有效的,具有一定的推广价值。
위료진일보제고SVM집성적범화능력,제출료기우Choquet모호적분적SVMs집성방법,종합고필각개자SVM수출중요성,피면료현유SVM집성방법중홀략차요신식적문제。응용해방법,이고교적구역경제공헌도위례진행방진시험,결과표명기우Choquet모호적분적SVMs집성방법교기우Sugeno모호적분SVMs집성방법화기우투표책략적SVMs집성방법구유경고적준학성。해방법시가행、유효적,구유일정적추엄개치。
In order to improve the classification performance of the support vector machine （SVM） ensemble methods, a modified SVM ensemble method is put forward by using Choquet fuzzy integral other than Sugeno integral. The proposed method takes the output of every SVM component into account such that it overcomes the drawback of the existing SVM ensemble methods that neglect the secondary information. As an example, based on the data collected in Shandong Province, the proposed method is used to evaluate the performance of social service made by the colleges in the Province. Simulation results show that the proposed method outperforms the existing SVM ensemble methods in the sense of classification performance.