论文部分内容阅读
研究了神经信号传输中, 内源性一氧化氮(NO)的四维动态扩散特性及其在长时程学习过程中的增强作用. 将NO扩散机理建模后与Kohonen自组织映射模型相结合, 在空间SOM基础上引入时间增强, 提出了新型的动态扩散型自组织映射模型, 计算了最优化权值的误差函数, 并分析比较了该模型与SOM中噪声干扰对训练的影响. 最后结合多项典型的一维和二维输入模式,给出自组织映射模型与扩散型自组织映射模型的仿真结果对比.
The four-dimensional dynamic diffusion of endogenous nitric oxide (NO) and its enhancement in long-term learning process were studied in neural signal transmission.With the combination of Kohonen self-organizing map model and NO diffusion mechanism modeling, The time-based SOM is introduced, a new model of dynamic diffusion self-organizing map is proposed, the error function of the optimal weight is calculated, and the impact of noise interference on SOM training is analyzed and compared. Finally, A typical one-dimensional and two-dimensional input mode is given, and the simulation results of the self-organizing map model and the diffusion self-organizing map model are given.