关于泛化神经网络与支持向量机的研究

人工神经网络(ANN)的泛化特性是神经网络最重要的特性,同时也是最不容易保证的特性。本文对改进泛化的神经网络算法以及新兴的机器学习算法——支持向量机算法进行研究,并分别用BP神经网络、改进泛化能力的神经网络、支持向量机对人体脂肪测试实例进行仿真预测分析,结果表明,支持向量机比神经网络、改进泛化神经网络具有更好的预测(泛化)能力,是人工神经网络的替代方法。

The research on Neural Networks with Enhanced Generalization and Support Vector Machine

The generalization ability of an artificial neural network(ANN) is the most important performance of it,but to obtain the good generalization capability of an ANN is not an easy thing.This paper introduces the main algorithms of the improved neural networks with enhanced generalization and the popular machine learning algorithm——Support Vector Machine(SVM),based on BP neural network,the improved neural networks with enhanced generalization and the SVM to simulate fat testing experiment.The result indccates that the generalization ability of SVM is stronger than the neural network and the improved neural networks with enhanced generalization.So the SVM is destined to be the substitute for the ANN.