基于聚类分析的综合神经网络集成算法

An Integrated Neural Network Ensemble Algorithm Based on Clustering Technology

研究神经网络集成是一种有效实用的分类方法,权值是影响神经网络集成性能的重要因素。为了克服神经网络集成固定权值的缺陷,提出一种基于聚类分析的综合神经网络集成算法。算法首先将样本分类,每类样本中加入其他样本类一定数量的中心样本,不同的神经网络学习不同类的样本。根据输入数据与样本类别之间的相关程度自适应调整集成权值。算法不仅用于自适应调整集成权值,而且是一种产生个体神经网络的训练方法。四个数据集上的仿真试验证实了算法的有效性。

Different component neural networks（NNs） in an ensemble in which different training sets have different performance for the same input data.The weights of an ensemble impact greatly on the performance of ensemble.The fixed weights may weaken the performance of some component NNs which can have better performance and lower weights,An Integrated neural network ensemble（InNNE） is proposed in the paper,which is an integrated ensemble algorithm not only for dynamically adjusting weights of an ensemble,but also for generating component NNs based on clustering technology.InNNE classifies the training set into different training subsets with clustering technology,which are used to train different component NNs.The weights of an ensemble are adjusted by the correlation of input data and the center of different training subsets.InNNE can increase the diversity of component NNs and decrease generalization error of ensemble.The paper provides both the analytical and experimental evidence to support the novel algorithm.