摘要
由于置信向量机运算大、分类速度慢导致其应用价值有限,需要对其进行改进。详细分析了近邻置信向量机所使用的基本技术,论述了近邻置信向量机使用的奇异检测函数和分类方法,并将其与基本置信向量机进行了对比。给出了近邻置信向量机的具体实施步骤。通过试验证明解决了置信向量机运算量大的问题,提高了分类速度。
The confidence support vector machine should be improved because of its much computing, slow classification speed and limited application value. This paper analyzes in detail basic technique used in nearest neighbor confidence support vector machine （NNCSVM）, and discusses oddity detecting function and classification method adopted in NNCSVM, compares NNCSVM to original confidence support vector machine, and presents typical implementation steps of NNCSVM. The experiment results show that this technique can decrease computing amount of confidence support vector machine and increase its classification speed.
出处
《计算机与网络》
2010年第3期83-85,共3页
Computer & Network
关键词
置信向量机
置信度
可靠性
近邻置信向量机
confidence support vector machine
confidence
reliability
nearest neighbor confidence support vector machine
