一种改进FHN神经元滤波模型研究

Research on an improved FHN neuronal filtering model

突触滤波是神经元处理和传递信息的重要过程,有助于生物在复杂环境中获取所需信息。针对当前人工神经元模型中较少考虑到突触滤波机制,本文以FitzHugh-Nagumo(FHN)人工神经元模型为基础构建基于膜电势增量变化的神经元数学模型,在此基础上模拟突触滤波机制,从而提出一种改进FHN神经元滤波模型。而后,对该模型的稳定性条件、幅频响应进行了分析,并通过不同信噪比条件下的典型信号和语音信号实验对该模型的信息传递能力和滤波能力进行验证。实验结果表明,该模型能够有效传递输入信息、提高输入信息强度,且有效抑制其中噪声部分。

Synaptic filtering,which is quite helpful to get the information needed in complex environment for living things,is an important process for neurons to process and transmit information.For synaptic filtering is rarely considered in modeling the artificial neuron models,this paper proposes an improved FitzHugh-Nagumo(FHN)model.By building a neuron model that can describe the incremental change of membrane potential based on the FHN model and simulating the synaptic filtering on this basis,the mathematical description of the proposed improved FHN model is derived.Then,the stability condition and the responses are discussed,and the information transmit ability and the filtering ability of the proposed model are tested by the typical signals and the speech signals in the cases of different conditions with different signal-noise ratios(SNRs).The experiments verify that,the model can realize the transmission of inputs,increase the intensity of inputs and reduce the noises of inputs effectively.