神经元参数估计的符号动力学方法

Neural Parameter Estimate with Symbolic Dynamics

提出一种与现有的参数估计完全不同的方法—用符号动力学实现参数估计。按符号动力学原理对系统输出信号进行粗粒化测量,根据符号序列的距离来衡量两条轨道的接近程度,可实现高精度的参数估计,精度与输出符号序列的长度相关。采用26个脉冲,估计的误差即可小于1/1000。该方法不需要对系统输出作精确测量,对不稳定系统也能实现估计,对于脉冲式工作的神经元的参数估计尤为有效。

In this paper,it was proposed that neural parameters were estimated by symbolic dynamics,which was different from the existing methods.Based on the principle of symbolic dynamics,the distance between symbolic sequences can be determined by measuring the degree of nearness between two orbits.The high precision estimation could be implemented without measuring the output in high precision.The error was less then 1/1000 when there were 26 pulses in the output spike sequence.The proposed method can be applied to unstable systems,and is especially efficient for parameter estimation of pulsed neuron.