Research On Stimulus Condition Information Of Signal Transmission In Neural Networks

Up to now, the physiology and anatomy of the nervous system has achieved important results, it has been known about the neuronal functions, operation mode and physiological structure from the cellular and molecular level in the nervous system rapidly. It is found that there are certain kinds of noises in both internal and external of the nervous system inevitably, and there is a synergy phenomenon induced by the noise in nervous systems.The thesis uses the stimulus-specific information to explore the performance of information transmission of the Integrate-and-Fire model. It is noted that the input speech signal and the noise such as Gaussian white noise and colored noise components are interactional in neuronal networks. Firstly, this thesis applies the average mutual information in the environment of white noise to explore the possibility of suprathreshold stochastic resonance. Then this thesis uses the stimulus-specific information measure to explore the coding efficiency of each component in the stimulus signal, and also evaluate their contributions to the average mutual information. The results showed that the stimulus-specific information distributed in the neural network converges on the amount of the average mutual information statistically. It is also shown that the stimulus-specific information can measure the coding efficiency in each component of the input signal via the internal noise enhancement clearly. The results also proved that adding a certain amount of colored noise can improve the transmission performance of the nervous system and the enhancing rate based on the frequency of colored noise. Besides, different intensity of colored noise has different effects on the distributed space of stimulus-specific information. These results have great significance in the coding scheme of neuronal integration models.