A precision-aware spike timing determination in Courbage-Nekorkin-Vdovin neuron map model: Statistical approaches and...

This study focuses on five different neural dynamics related to spike timing determination in the Courbage-Nekorkin-Vdovin (CNV) neuron map model. These neural dynamics are simulated numerically through the double-precision arithmetic and fixed-point arithmetic, separately. Notable temporal shifts are observed between the spike timings generated by these simulations. Since the accurate determinations of the spike timings are the critical requirement for modeling the functional behavior of neurons, three different statistical methods are used to determine the spike timings of these neuron behaviors in these performed numerical simulations. The concordances between the spike timings of the systems, which are simulated through two separate arithmetic approaches, are compared by calculating the means and the standard deviations of their time differences. This process is repeated for three statistical methods, namely, "mean + K × Standard Deviation (Std)", "median + K × Median Absolute Deviation (MAD)", and "z-score normalization" methods. On the other hand, the requirement to fixed-point arithmetic usage arises due to the limited resources and energy consumption of the digital hardware. Accordingly, in addition to presenting a statistical analysis study that considers the spike timing conditions during the simulations of the CNV neuron map model, this neuron map model is constructed on the Field Programmable Gate Array (FPGA) device by using the fixed-point arithmetic. Therefore, this study aims additionally to verify the applicability of the process on hardware by performing its electronic implementation.