Parameter estimation for coupling neural network models with symbolic dynamics

Abstract

This paper presents work on parameter estimation method for simple coupling neural network models. Different from the traditional voltage-clamp technique to extract each ion channel parameters of a neuron, the method proposed in this paper only need to record the inter-spike interval sequences of the neuron's output. Based on the principle of symbolic dynamics, the action potential sequences can be symbolized without high precision measurement. By computing the distance between symbolic sequences can analyze the degree of nearness between the two orbits, and then use dichotomy to find the optimal parameters. The longer the output spike sequence is, the higher precision estimation can be achieved. The proposed method is efficient for parameter estimation in unstable neural systems, and has a certain reference value for creating neural models from neural electrophysiological experiments.