Analog VLSI implementation of resonate-and-fire neuron

Area of Science:

• Neuroscience
• Analog Circuit Design
• Computational Neuroscience

Background:

• Spiking neuron models are crucial for understanding neural computation.
• Resonate-and-fire neuron (RFN) models exhibit complex dynamics like coincidence detection.
• Existing models may face challenges in large-scale analog implementation.

Purpose of the Study:

• To propose and validate an analog integrated circuit for the resonate-and-fire neuron (RFN) model.
• To implement the RFN model based on the Lotka-Volterra (LV) system in hardware.
• To demonstrate the circuit's functionality as a coincidence detector and band-pass filter.

Main Methods:

• Derivation of an analog circuit from the Lotka-Volterra (LV) system to mimic RFN dynamics.
• Utilizing circuit simulations to analyze the proposed RFN circuit's behavior.
• Evaluating circuit performance under conditions of additive white noise and background random activity.

Main Results:

• The proposed analog circuit successfully implements the resonate-and-fire neuron (RFN) model.
• Circuit simulations confirmed the RFN circuit's capability for coincidence detection.
• The circuit demonstrated band-pass filtering characteristics at the circuit level.
• Robust performance was observed even with added noise and random background activity.

Conclusions:

• The developed analog RFN circuit effectively mimics the dynamical behaviors of the mathematical model.
• The circuit's ability to function as a coincidence detector and band-pass filter is validated.
• This work is expected to facilitate very large-scale integration (VLSI) of functional spiking neural networks.