欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
支持向量机一对一多类分类在测试阶段需执行在训练阶段构造好的所有子分类器，耗费较长时间，这一缺点极大地限制了其在大规模数据识别中的应用。提出一种改进的一对一多类分类方法，在测试的中期阶段先对所有类别的得票数进行统计，将得票较低者剔除，不必计算由这些类别构成的子分类器，从而有效减少子分类器的数目。最后将此改进方法应用到抗噪语音识别系统中，实验结果表明该方法具有一定的优势。
英文摘要：
For one-against-one multi-classification method of Support Vector Machine （SVM）, all sub-classifiers that have been constructed during training phase are executed, which takes longer predicting time. This shortcoming greatly limits its application in the identification of large-scale datasets. Hence an improved one-against-one multi-class support vector machine is proposed. In the middle of the testing phase, statistics votes of all the catego- ries,the lower of which will be removed, and the sub-classifiers constituted by these categories is not caculated. In this case, sub-classifiers can be reduced effectively. Finally, the improved method is applied to speech recognition system with noise immunity. The experiment results show that the method has certain advantages.
同期刊论文项目
同项目期刊论文
Parameter optimization and application of support vector machine based on parallel artificial fish s
Study for classification of emotional speech by using optimized frame zero crossing with peak amplit
A robust feature extraction approach based on an auditory model for classification of speech and exp
期刊信息
位置：