【作者】 叶德谦； 刘波；

【Author】 YE De-qian LIU Bo ICDZ - Institute for Information Technology, Yanshan University, Qin Huangdao 066004, China

【机构】 燕山大学中德信息技术合作研究所；

【摘要】 由于BP神经网络有自身的很多缺陷,为改善其网络的性能,将它与一种改进的遗传算法相结合,这种具有性别特征的遗传算法根据性别的不同而进行不同的交叉变异操作,这能有效地保持群体的多样性,避免算法陷入局部极值而出现早熟现象,并将其用于对股市的预测,取得了较好的预测效果。更多还原

【Abstract】 Because BP neutral network has some defect, we combine neutral network with improved genetic algorithm in order to improve the performance of the BP neutral network. This genetic algorithm which has the character of sex could do different operation of crossover and mutation according to the different sex. So it can preserve the diversity of population effectively and avoid falling into partial extremum and the phenomena of prematurity. We finally apply it to prediction of stock market and have obtained good effect.更多还原