Common-mode noise suppression via differential readout in paired Rulkov neurons

Area of Science:

• Computational neuroscience
• Neuromorphic engineering
• Nonlinear dynamics

Background:

• Shared background noise hinders signal extraction in biological neural systems and neuromorphic circuits.
• Differential readout strategies are explored to mitigate noise effects.

Purpose of the Study:

• To investigate the efficacy of an opponent-channel differential readout using two parameter-matched Rulkov neurons in suppressing common-mode noise.
• To analyze the mechanisms underlying noise reduction and its limitations in this differential readout system.

Main Methods:

• Utilized two parameter-matched Rulkov neurons driven by opposite-polarity inputs and common-mode noise.
• Tested various stimuli including exponential-decay, alpha-function, step, and sinusoidal inputs.
• Performed mechanistic analysis to understand noise leakage in the nonlinear dynamics.

Main Results:

• The differential readout significantly improved waveform preservation and signal retention compared to single-neuron and amplitude-matched controls, especially under substantial common-mode noise.
• The performance advantage diminished with weak noise correlation or significant parameter mismatch.
• Common-mode noise was found to perturb the common operating state and re-enter the differential channel via state-dependent gain and local slope mismatch.

Conclusions:

• Opponent-channel differential readout in paired Rulkov neurons offers a robust method for enhancing signal extraction in the presence of shared noise.
• Nonlinear dynamics introduce leakage, limiting the complete elimination of common-mode noise and defining the boundaries of this readout strategy.