Dynamical system design for silicon neurons using phase reduction approach.

Area of Science:

• Neuroscience
• Computer Engineering
• Dynamical Systems Theory

Background:

• Silicon neurons (SiNs) are crucial for developing neuromorphic computing hardware.
• Existing SiN designs require efficient methods for parameter tuning and network synchronization.
• Phase reduction theory offers a powerful framework for analyzing and controlling oscillatory systems.

Purpose of the Study:

• To apply a computer-aided phase reduction approach for dynamical system design of SiNs.
• To clarify design criteria for SiNs based on phase response properties.
• To demonstrate delay-induced synchronization in a network of custom SiNs.

Main Methods:

• Reviewing dynamical system design principles for SiNs.
• Summarizing phase response properties of existing SiN circuit models.
• Applying phase reduction theory to tune circuit parameters of the resonate-and-fire neuron (RFN) circuit.
• Simulating a silicon spiking neural network composed of RFN circuits.

Main Results:

• Successful application of computer-aided phase reduction to SiN design.
• Identification of key design criteria based on phase response properties.
• Demonstration of delay-induced synchronization in a network of hybrid SiNs (RFN circuits).

Conclusions:

• The computer-aided phase reduction approach is effective for SiN dynamical system design.
• Parameter tuning of RFN circuits enables controlled synchronization in SiN networks.
• This method facilitates the development of advanced neuromorphic computing systems.