A Novel Asynchronous CA Neuron Model: Design of Neuron-Like Nonlinear Responses Based on Novel Bifurcation Theory of Asynchronous Sequential Logic Circuit
A Novel Asynchronous CA Neuron Model: Design of Neuron-Like Nonlinear Responses Based on Novel Bifurcation Theory of Asynchronous Sequential Logic Circuit
复制标题
DOI：
10.1109/tcsi.2020.2971786
复制
发表时间：
2020-02
期刊：
影响因子：
--
通讯作者：
Kentaro Takeda;H. Torikai
中科院分区：
文献类型：
--
作者：
Kentaro Takeda;H. Torikai
提出了一种新的异步时序逻辑电路神经元模型，该模型的非线性动力学用异步元胞自动机来描述。作为一种新的理论分析工具，本文导出了连续离散混合庞加莱映射。该图谱显示，所提出的模型可以再现生物学上可信的神经元模型的基本动态特性和典型分岔机制。根据分岔分析结果，提出了该模型的系统设计方法。结果表明，所提出的设计方法能够再现典型的神经元非线性响应曲线及其潜在的分岔机制。最后，在现场可编程门阵列中实现了该模型，并进行了实验验证。结果表明，与传统的神经元模型相比，该模型可以用更少的电路元件实现，并且功耗更低。
This paper presents a novel asynchronous sequential logic circuit neuron model, the nonlinear dynamics of which is described by an asynchronous cellular automaton. As a novel theoretical analysis tool for the model, a continuous-discrete hybrid Poincare map is derived. The map reveals that the presented model can reproduce fundamental dynamic properties and typical bifurcation mechanisms of a biologically plausible neuron model. Using the bifurcation analysis results, systematic design methods of the presented model are proposed. It is shown that the design methods enable the presented model to reproduce typical nonlinear response curves of neurons as well as their underlying bifurcation mechanisms. In addition, the model is implemented in a field programmable gate array and experiments validate its operation. It is shown that the presented model can be implemented by fewer circuit elements and consumes lower power compared to typical conventional neuron models.