Protocol Classification Algorithm of CK-means Based on Dense Distance
To address the problems of class unevenness and center point selection in protocol classification,a Can-opy protocol classification algorithm based on maximum density distance is proposed.The algorithm uses samples to determine the distance parameters of the algorithm without manual setting,determines the center point and eliminates outliers in the clustering process by calculating the sample density value.The maximum dense distance is applicable to non-uniform data sets,and the calculated center point is closer to the center of the category,which can improve the stability and reliability of clustering.Experimental analysis shows that,compared with other algorithms,the optimized algorithm can overcome the adverse effects of noise and class imbalance on protocol identification,and improve the clustering effect while improving the accuracy of protocol identification.
Distance parametersMaximum density distanceProtocol classification
NETL
NSTL
万方数据
