摘要
在无线电频谱监测应用中,要求能够及时发现观测频段内频谱图像的异常情况。提出了一种基于单类支持向量机(v-OCSVM)模型的频谱异常检测方法,通过实验检测和ROC曲线分析,表明v-OCSVM模型用于频谱异常检测是可行的。
In applications of the radio spectrum monitoring, it requires to be found in the frequency spectrum in time observation abnormalities of the image. This paper puts forward anomaly detection methods in spectrum based on one class support vector machine (v -OCSVM) model, through the experimental detection and ROC curves analysis, shows that v - OCSVM model can bu used for spectrum anomaly detection.
出处
《微处理机》
2012年第4期73-75,79,共4页
Microprocessors
关键词
频谱检测
单类支持向量机
ROC曲线
Spectrum testing
One class support vector machine
ROE curves