CAJ | 학술논문

为了提高网络热点事件的预测精度，提出一种基于遗传优化加权相关向量机的网络热点事件预测模型（GA-WRVM）。首先对每个样本数据的噪声方差加上一个权值系数得到加权相关向量机，然后用组合核函数代替相关向量机的单一核函数，并采用遗传算法对核参数进行优化建立预测模型，最后利用网络热点数据进行仿真实验分析。结果表明，相对于神经网络和支持向量机， GA-WRVM降低了网络热点事件的预测误差，获得了更高的网络热点预测精度，具有更强的泛化能力。
위료제고망락열점사건적예측정도，제출일충기우유전우화가권상관향량궤적망락열점사건예측모형（GA-WRVM）。수선대매개양본수거적조성방차가상일개권치계수득도가권상관향량궤，연후용조합핵함수대체상관향량궤적단일핵함수，병채용유전산법대핵삼수진행우화건립예측모형，최후이용망락열점수거진행방진실험분석。결과표명，상대우신경망락화지지향량궤， GA-WRVM강저료망락열점사건적예측오차，획득료경고적망락열점예측정도，구유경강적범화능력。
In order to improve the prediction accuracy of network hotspot events,in this paper we propose a new network hotspots events prediction model,which is based on using genetic algorithm to optimise weighted relevance vector machine (GA-WRVM).First,we impose the weighting coefficient to the noise variance of every sample data to get the weighted relevance vector machine (WRVM).Then we use the combined kernel function to take place the single kernel function of WRVM,and employ genetic algorithm to optimise kernel function parame-ters to build the prediction model.Finally,we carry out the simulation experiment by using network hotspots event data.Results show that compared with the neural network and support vector machine,GA-WRVM reduces the prediction errors of network hotspot events,and obtains higher prediction accuracy,thus has stronger generalisation ability.