摘要
简要分析由Logistic映射产生的混沌数以及不同混沌序列之间的概率统计特性,为混沌全局性搜索提供了依据.将一种快速BP算法与混沌优化相结合,提出了混沌BP混合算法.由于混沌Logistic映射的遍历性、随机性,使得混合算法收敛速度快,且具有全局性.采用混合算法对XOR问题和非线性函数进行仿真,结果表明该算法明显优于标准BP算法和快速BP算法.
Probabilistic properties are analyzed for chaotic data and different chaotic sequences generated by Logistic map, which provides theoretical basis for chaos global searching. A chaos-BP hybrid algorithm is proposed by means of combination of a new fast BP algorithm and chaos optimization searching. Due to ergodicity and random of chaotic Logistic map, chaos-BP algorithm converges fast and globally, and has no local minimum. The algorithm is applied to XOR problem and nonlinear function approximation. Simulation results show that the chaos-BP algorithm needs shorter learning time than that of the standard BP and fast BP.
出处
《控制与决策》
EI
CSCD
北大核心
2004年第4期462-464,共3页
Control and Decision
基金
湖南省自然科学基金资助项目(01JJY3029).
关键词
前馈神经网络
混沌优化
BP算法
遍历性
feedforward neural network
chaos optimization
BP algorithm
ergodicity