CAJ | 학술논문

为了提高复杂多变的网络流量预测精度，提出了一种基于仿射传播聚类算法和稀疏贝叶斯的网络流量预测模型。采用仿射传播聚类算法对网络流量训练集进行聚类，从而将网络流量训练集划分为若干个子类，然后采用稀疏贝叶斯回归为每个子类建立相应的预测模型，最后采用具体的网络流量数据对模型的性能进行测试。实验结果表明，模型可以获得比较理想的网络流量预测结果，预测误差可以满足网络流量的实际应用要求。
위료제고복잡다변적망락류량예측정도，제출료일충기우방사전파취류산법화희소패협사적망락류량예측모형。채용방사전파취류산법대망락류량훈련집진행취류，종이장망락류량훈련집화분위약간개자류，연후채용희소패협사회귀위매개자류건립상응적예측모형，최후채용구체적망락류량수거대모형적성능진행측시。실험결과표명，모형가이획득비교이상적망락류량예측결과，예측오차가이만족망락류량적실제응용요구。
In order to improve the prediction accuracy of complex network traffic,this paper proposed a novel network traffic prediction model based on affinity propagation clustering algorithm and sparse Bayesian.Firstly,it used the affinity propagation clustering algorithm to cluster the network traffic training set,to divided the network traffic training set into several sub catego-ries.Then it used the sparse Bayesian regression to establish prediction models for each sub categories.Finally it tested the per-formance of network traffic prediction model on specific network traffic data to the model.The experimental results show that the proposed model can obtain more ideal predict results of network traffic,the prediction error can satisfy the practical applica-tion requirement of network flow.