函数链网络(Functional Link Network——FLN)通过对输入向量(或模式)的非线性扩展,将非线性映射特性引入了单层神经网络,采用δ学习规则获得了快速的学习和非线性映射特性。本文在FLN基础上,借助凸集优化思想,利用最陡梯度下降技术获...函数链网络(Functional Link Network——FLN)通过对输入向量(或模式)的非线性扩展,将非线性映射特性引入了单层神经网络,采用δ学习规则获得了快速的学习和非线性映射特性。本文在FLN基础上,借助凸集优化思想,利用最陡梯度下降技术获得了比FLN更高的存储容量和更快速的学习速度。计算机模拟的结果证实了所提的算法性能。展开更多
Study of chaotic synchronization as a fundamental phenomenon in the nonlinear dynamical systems theory has been recently raised many interests in science, engineering, and technology. In this paper, we develop a new m...Study of chaotic synchronization as a fundamental phenomenon in the nonlinear dynamical systems theory has been recently raised many interests in science, engineering, and technology. In this paper, we develop a new mathematical framework in study of chaotic synchronization of discrete-time dynamical systems. In the novel drive-response discrete-time dynamical system which has been coupled using convex link function, we introduce a synchronization threshold which passes that makes the drive-response system lose complete coupling and synchronized behaviors. We provide the application of this type of coupling in synchronized cycles of well-known Ricker model. This model displays a rich cascade of complex dynamics from stable fixed point and cascade of period-doubling bifurcation to chaos. We also numerically verify the effectiveness of the proposed scheme and demonstrate how this type of coupling makes this chaotic system and its corresponding coupled system starting from different initial conditions, quickly get synchronized.展开更多
文摘函数链网络(Functional Link Network——FLN)通过对输入向量(或模式)的非线性扩展,将非线性映射特性引入了单层神经网络,采用δ学习规则获得了快速的学习和非线性映射特性。本文在FLN基础上,借助凸集优化思想,利用最陡梯度下降技术获得了比FLN更高的存储容量和更快速的学习速度。计算机模拟的结果证实了所提的算法性能。
文摘Study of chaotic synchronization as a fundamental phenomenon in the nonlinear dynamical systems theory has been recently raised many interests in science, engineering, and technology. In this paper, we develop a new mathematical framework in study of chaotic synchronization of discrete-time dynamical systems. In the novel drive-response discrete-time dynamical system which has been coupled using convex link function, we introduce a synchronization threshold which passes that makes the drive-response system lose complete coupling and synchronized behaviors. We provide the application of this type of coupling in synchronized cycles of well-known Ricker model. This model displays a rich cascade of complex dynamics from stable fixed point and cascade of period-doubling bifurcation to chaos. We also numerically verify the effectiveness of the proposed scheme and demonstrate how this type of coupling makes this chaotic system and its corresponding coupled system starting from different initial conditions, quickly get synchronized.