CAJ | 학술논문

样条权函数神经网络是一种新兴的神经网络，克服了很多传统神经网络(如BP、RBF)的缺点：比如局部极小、收敛速度慢等。它具有拓扑结构简单，精确记忆训练过的样本，反映样本的信息特征，求得全局最小值等优点。基于这些优点，文中提出了一种基于样条权函数神经网络P2 P流量识别方法。通过提取P2 P流量特征，运用样条权函数神经网络结构对P2P流识别。 Matlab仿真和模拟实验结果表明了这种方案的可行性，与传统神经网络相比，样条权函数神经网络在时间效率上具有明显优势。
양조권함수신경망락시일충신흥적신경망락，극복료흔다전통신경망락(여BP、RBF)적결점：비여국부겁소、수렴속도만등。타구유탁복결구간단，정학기억훈련과적양본，반영양본적신식특정，구득전국최소치등우점。기우저사우점，문중제출료일충기우양조권함수신경망락P2 P류량식별방법。통과제취P2 P류량특정，운용양조권함수신경망락결구대P2P류식별。 Matlab방진화모의실험결과표명료저충방안적가행성，여전통신경망락상비，양조권함수신경망락재시간효솔상구유명현우세。
Spline weight function neural network is a new kind of neural network. It overcomes many defects of traditional neural networks ( like BP,RBF) ,such as local minima,slow convergence,at the same time has many advantages,such as simple structure,remembering trained samples,reflecting the characteristics of the sample information,finding global minima directly and so on. A method of P2P traffic identification based on spline weight function neural network is presented in this paper based on advantages of this neural network. The structure of spline weight function neural network can identify P2P traffic by extracting characteristics of P2P traffic training. Matlab sim-ulation and experimental results show the feasibility of the scheme. Compared with the traditional neural network,spline weight function neural network has obvious advantages in time efficiency.