Research On Neural Information Chaos Coding Mechanism And Its Application

In the process of neural information transmission,a growing number of experiments and simulations have revealed the presence of chaotic rhythm.To study the chaotic coding and transmission mechanism,will help to understand the closely relationship between the neural coding and neural function.Furthermore,it will bring new ideas to the engineering applications,such as neural computing and artificial intelligence.First of all,on the basis of neuronal electrophysiological model,we studied the patterns of information processing for chaotic spike train.And we proposed a new mechanism of chaotic neuronal information coding,which was applied to the measurement of weak signal.In addition,by using dynamic synapses,we investigated the neural information interaction under chaotic forcing and constructed a visual neural computing model,and then a new ideas for detecting weak image edges based on the chaotic coding mechanism was proposed.At the end of this thesis,taking advantage of the ergodicity and irreversible characteristic of chaotic response,we proposed an application of identity authentication system based on two dimensional chaotic image.The main results of this thesis are summarized as follows:(1)For the chaotic states of the neuron under periodic forcing,we recognized the spike patterns by using the ISI sequences and nonlinear prediction methods;Considering the statistical limitations of traditional temporal code or rate code,a chaotic neuronal information coding mechanism was proposed based on circle map;By implementing a coarse-grained algorithm—symbolic dynamics,the spike trains were transformed into symbolic sequences.Based on the ordering rules of symbolic sequences,we proposed a measurement methods of weak signal.The experimental results showed that the chaotic encoding mechanism would enhance neuronal sensitivity to external stimulus,and were conducive to improve the efficiency of neurotransmission.(2)For the characteristic of chaotic neural information transmission,we studied the dynamic states of neurons under the chaotic forcing,by using the approximate entropy and the nonlinear prediction method.Experimental results showed that in the chaotic information transmission,the postsynaptic neurons significantly retained the excitation signal characteristics with the frequency alternating vary greatly,while discarded excitation signal characteristics with frequency alternating slightly,which indicated that neuronal chaotic information transmission had selectivity and reprocessing capabilities,and was in favor of information interaction in the group coding.(3)Based on the mechanism of dynamic neuronal synaptic transmission and chaos coding,we constructed a new visual neural computing model and tried to detect weak edge of image.According to changing the signal conversion ability of postsynaptic neuron by the pattern of spike train,we took advantage of the postsynaptic membrane potential waveform to depict the spatial and temporal characteristics of image.Experimental results showed that the new model would reflect the state of temporal dynamics of neural network evolution,converting optical signals to spike sequence with spatial and temporal characteristics and to improve the sensitivity of the weak edge detection by using the chaotic coding mechanism.(4)Based on chaotic information coding characteristics,we proposed a new method for image password configuration.Taking advantage of nonlinear mapping,we transformed user ID and digital password into dynamic chaotic sequence,and then drove neurons to produce irreversible and chaotic spike train.Finally we reconstructed the spike train into a two-dimensional image password.The new method effectively enhanced the security of password authentication,and provided a new way of thinking for password encryption and authentication field application.