基于个体选择的动态权重神经网络集成方法研究

神经网络集成技术能有效地提高神经网络的预测精度和泛化能力,已成为机器学习和神经计算领域的一个研究热点。该文针对回归分析问题提出了一种结合应用遗传算法进行个体选择和动态确定结果合成权重的神经网络集成构造方法。在训练出个体神经网络之后,应用遗传算法对个体网络进行选择,然后根据被选择的各个体网络在输入空间上对训练样本的预测误差,应用广义回归网络来动态地确定各个体网络在特定输入空间上的合成权重。实验结果表明,与仅应用个体网络选择或动态确定权重的方法相比,该集成方法基本上能取得更好地预测精度和相近的稳定性。

Dynamically Weighted Ensemble Neural Networks Which Selected by Genetic Algorithm for Solving Regression Problems

Combining the outputs of several neural networks into an aggregate output often gives improved accuracy over any individual output.The set of networks is known as an ensemble.This paper presents an ensemble method for regression that has advantages over usual ensemble techniques.Generally,the approach ensembles all component networks for prediction,and the output of an ensemble is a weighted sum whose weights are fixed.This ensemble approach select some component networks by genetic algorithm firstly,and the combining weights of the component networks are dynamic,the weights dynamically determined from the predicted accuracies of the trained networks with training dataset,The more accurate a network seems to be of its prediction,the higher the weight.This is implemented by generalized regression Neural Network.Empirical results show that this method improved on prediction accuracy and stability.