现代电子技术
現代電子技術
현대전자기술
Modern Electronics Technique
2015年
21期
109-112,117
,共5页
有督导机器学习%网络流量识别%LSSVM%协同量子粒子群优化算法
有督導機器學習%網絡流量識彆%LSSVM%協同量子粒子群優化算法
유독도궤기학습%망락류량식별%LSSVM%협동양자입자군우화산법
supervised machine learning%network traffic identification%LSSVM%CQPSO algorithm
针对真实网络环境中存在大量干扰噪声和野值样本等严重影响最小二乘支持向量机算法的性能等问题,提出一种结合协同量子粒子群优化算法和最小二乘支持向量机的网络流量识别系统.将网络流量分为12个类型,并进行数据采集.使用采集的数据对网络流量识别系统进行训练和性能测试.为研究提出的基于CQPSO-LSSVM算法的性能,将其与基于CQPSO-LSSVM算法和基于PSO-LSSVM算法进行对比,结果表明基于CQPSO-LSSVM算法具有更快的识别速度以及更好的识别准确率,避免了出现陷入局部最优解的情况发生.
針對真實網絡環境中存在大量榦擾譟聲和野值樣本等嚴重影響最小二乘支持嚮量機算法的性能等問題,提齣一種結閤協同量子粒子群優化算法和最小二乘支持嚮量機的網絡流量識彆繫統.將網絡流量分為12箇類型,併進行數據採集.使用採集的數據對網絡流量識彆繫統進行訓練和性能測試.為研究提齣的基于CQPSO-LSSVM算法的性能,將其與基于CQPSO-LSSVM算法和基于PSO-LSSVM算法進行對比,結果錶明基于CQPSO-LSSVM算法具有更快的識彆速度以及更好的識彆準確率,避免瞭齣現陷入跼部最優解的情況髮生.
침대진실망락배경중존재대량간우조성화야치양본등엄중영향최소이승지지향량궤산법적성능등문제,제출일충결합협동양자입자군우화산법화최소이승지지향량궤적망락류량식별계통.장망락류량분위12개류형,병진행수거채집.사용채집적수거대망락류량식별계통진행훈련화성능측시.위연구제출적기우CQPSO-LSSVM산법적성능,장기여기우CQPSO-LSSVM산법화기우PSO-LSSVM산법진행대비,결과표명기우CQPSO-LSSVM산법구유경쾌적식별속도이급경호적식별준학솔,피면료출현함입국부최우해적정황발생.
In the real network environment,a large number of interference noise and outlier samples are existed,which se-riously affect on the performance of the least square support vector machine(LSSVM)algorithm. A network traffic identification system combining cooperative quantum particle swarm optimization (CQPSO) algorithm with LSSVM is proposed. The network traffic is divided into 12 types,in which the data of network traffic are collected. The network traffic identification system is con-ducted with training and performance test by the collected data. To study the performance of the CQPSO-LSSVM based algo-rithm,the CQPSO-LSSVM based algorithm is compared with the PSO-LSSVM based algorithm. The comparison results show that the CQPSO-LSSVM based algorithm has faster identification speed and better identification accuracy,which can avoid the occur-rence that the system is caught in local optimal solution.
