Design and Implementation of Neural Network Based Chaotic System Model for the Dynamical Control of Brain Stimulation

摘要

Brain stimulation has been used in practice to treat neurological diseases, such as Parkinson's Disease and Epilepsy. However, the stimulation signals are generated based on trail and error; and the underpinning theory of this treatment is still unclear. Artificial neural network (ANN) resembles biological neural network in the brain and has been used for many artificial intelligence applications such as classification and pattern recognition. In order to generate accurate stimulation signals in brain stimulation treatment, it is beneficial to establish an ANN model to simulate the brain dynamics and study the effects of various stimulation signals. Previous research shows that brain activities captured by Electroencephalogram (EEG) demonstrate chaotic patterns. Chaotic systems, such as Henon map can be represented by a set of mathematical equations, and therefore are predictable and controllable. The aim of this research is to implement an optimal ANN architecture model to generate the output pattern of a chaotic system, which can be used to simulate the brain dynamics under stimulation. This paper presented the preliminary work of an ANN architecture design and optimization for generating the outputs of Henon map chaotic system, and the simulation results for controlling the chaotic system with periodic stimulation signals. The ANN design method and chaotic control method can be extended for other chaotic systems in general.