The Algorithm Complexity Research Of Cubic Spline Weight Function Neural Network And Its Application

Different algorithm may be with a different time, space or efficiency to accomplish the sametask. For well-implemented algorithm can often bring higher operating and storage efficiency. Sostudy the complexity of the cubic spline weight function algorithm is very meaningful and goodalgorithms can help us be more effective to solve the problem.In order to study algorithm complexity of the cubic spline weight function neural network,this paper is based on the spline weight function neural network theory and combined with theknowledge of numerical analysis, algorithm to do the theoretical a nalysis and formula derivation tosolve the algorithm complexity of cubic spline weight function neural network. At the same time,on the basis of theoretical analysis, use MATLAB tools to simulate the algorithm complexity of thecubic spline weight function neural network.This paper ultimately concludes that the algorithm time complexity of the cubic spline weightfunction neural network has relation to sample input dimensions and output dimensions as well asthe number of sample points. When the input, output dimension are fixed, with the increase of thenumber of training samples point, the simulation time and number of samples are are linear growth.Through the analysis of simulation results to verify the correctness of theoretical analysis andthe cubic spline weight function neural network has the advantages of low algorithm complexityand fast training speed.On the basis of the analysis of the cubic spline weight function neural network algorithmcomplexity, this paper presents the application of the cubic spline weight function neural networkalgorithm in medical data mining for classification of diabetes. Use MATLAB tools to simulate theapplications of the cubic spline weight function neural network and BP neural network. Through theresults of the experiments, we know the cubic spline weight function neural network algorithm hasa good predictor of diabetes testing, training fast and simple structure of the classifier. Comparedwith the traditional BP neural network, it has advantages of training speed, high precision.