论文部分内容阅读
径向基函数网络(Radial Basis Function Network,RBFN)是二十世纪八十年代末提出的一种神经网络.当网络的输入维数较大时,RBFN的系统复杂性大大提高,从而使RBFN的行为受到影响,因此降低RBFN输入维数已成为RBFN的研究热点.本文提出一类基于RBFN的分工协作系统及其学习算法(A Divide-and-Cooperate Hybrid System Based RBFN,DCRBFN).DCRBFN是一种由多个子RBFN组成的混合结构,每个子RBFN具有自己的输入空间.由于DCRBFN把高维模型分解为低维模型,所以DCRBFN不仅明显降低了RBFN的复杂性而且网络的收敛速度更快.实验表明,DCRBFN在处理高维模型的行为明显优于RBFN.
Radial Basis Function Network (RBFN) is a neural network proposed in the late 1980s. When the input dimension of the network is large, the system complexity of RBFN is greatly increased, so RBFN So the reduction of the input dimension of RBFN has become a research focus of RBFN.This paper presents a class of division and cooperation based on RBFN and learning algorithm (A Divide-and-Cooperate Hybrid System Based RBFN, DCRBFN) .DCRBFN is a Due to the fact that DCRBFN decomposes the high-dimensional model into a low-dimensional model, DCRBFN not only reduces the complexity of RBFN but also speeds up the convergence of the network.Experiments show that the RBFN of sub-RBFN has its own input space, , DCRBFN in the treatment of high-dimensional model was significantly better than RBFN.