摘要
根据生物神经元是利用噪声通过随机共振机制进行信息处理的现象,以阀值神经元为例,采用香农第二定理,分析了双极性信号和白噪声通过由多个神经元串联或并联组成的系统的输出情况,从品质因数、传输长度等角度对多个神经元串联系统的输出特性随噪声强度变化进行了分析.实验结果表明:系统存在随机共振现象,即存在最优的噪声量可以最大限度地增强信息传输;串联时传输信息量随级数的增大而减小,并联时正好相反;噪声的分布形式只影响共振效应的强弱,而不影响系统的随机共振特性;均匀白噪声比高斯白噪声有更强的共振效应.从而说明了噪声对信息传输的积极作用.
According to the phenomenon that biological neurons process information by the utilization of noise via stochastic resonance （ SR）, the output of bipolar random signal and white noise through threshold neurons connected in series or in parallel were studied by Shannon＇s second theorem. The influence of noise distribution on SR was investigated and the characteristics of the neurons in series were analyzed in terms of goodness and propagation length. The results demonstrate the optimal noise-assisted information transmission due to SR. To the neural network in series, information transmission decreases as the levels increase, but it is quite the contrary to the parallel case. The noise variance only affects the intensity of SR, no matter with the SR phenomenon. Compared with Gaussian noise, the even white noise has stronger SR behavior. All these proved the noise was effective in signal transmission.
出处
《东南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2005年第5期710-713,共4页
Journal of Southeast University：Natural Science Edition
基金
国家自然科学基金资助项目(60475034)
教育部霍英东教育基金资助项目
江苏省自然科学基金资助项目(BK2001402)
关键词
随机共振
阈值神经元
信息理论
双极性信号传输
stochastic resonance
threshold neurons
information theory
bipolar signal transmission
