基于改进粒子群算法的神经网络优化证券投资组合方法

采用人工神经网络对证券投资进行预测与分析的研究过程中，提高神经网络各个节点参数的优化能力是极其关键的。传统的神经网络存在学习速度慢、易陷入局部极小值、预测结果精度较低等缺点，一种玫进型粒子群（Improved Particle Swann Optimizer，IPSO）算法．可以优化BP（Back Propagation）神经网络．并将优化后的BP神经网络应用于优化证券投资组合中。实验结果表明：该研究方法能够在预测精度和稳定性方面明显优于传统的PSO—BP神经网络优化证券投资组合方法。

The Optimization of Portfolio Method of Neural Network Based on Improved Particle Swarm Optimizer algorithm

In the artificial-neural- work-base forecast and analyses of portfolio investment, it is extremely important to improve the optimizing ability of each node parameters in neural network. Targeted to deal with traditional neural networks＇ shortcomings of slow learning speed, easy occurrence of local minimal value and lower prediction accuracy, we put forward an Improved Particle Swarm algorithm （Improved Particle Swarm Optimizer, IPSO）, with which BP （Back Propagation） neural network is optimized and applied to optimal securities portfolio. The results show that this method is superior than traditional PSO-BP neural network optimization portfolio method.