Neuromorphic System Based on CMOS Inverters and Si-Based Synaptic Device.

Area of Science:

• Neuromorphic Engineering
• Artificial Intelligence Hardware
• Solid-State Circuits

Background:

• Biological neurons exhibit complex functions like excitation, inhibition, and plasticity.
• Emulating these functions in artificial circuits is crucial for advancing neuromorphic computing.
• Existing artificial neuron circuits often require numerous components and high power consumption.

Purpose of the Study:

• To develop a compact and energy-efficient analog neuron circuit.
• To integrate silicon-based synaptic devices for neural computation.
• To implement key neuronal mechanisms including action potential generation and synaptic plasticity.

Main Methods:

• Designed an analog neuron circuit using n-channel and p-channel silicon-based synaptic devices.
• Incorporated current mirrors for synaptic connection and integration, mimicking biological excitation and inhibition.
• Utilized normal and modified inverters for action potential generation and implemented spike-timing-dependent plasticity (STDP) by connecting output potential to synaptic devices.

Main Results:

• Successfully demonstrated an analog neuron circuit capable of emulating biological neuron operations.
• Achieved excitation and inhibition mechanisms using synaptic devices and current mirrors.
• Implemented STDP for synaptic plasticity by adjusting synapse conductance.

Conclusions:

• The developed analog neuron circuit effectively emulates biological neuron functions using minimal devices and low power dissipation.
• The circuit's architecture, based on 4-terminal synaptic devices, eliminates the need for additional switches or logic operations.
• This approach offers a promising pathway for efficient and scalable neuromorphic hardware.