Simultaneous emulation of synaptic and intrinsic plasticity using a memristive synapse

Area of Science:

• Neuromorphic Engineering
• Materials Science
• Computational Neuroscience

Background:

• Neuromorphic computing aims to replicate brain functionality in hardware.
• Memristors have enabled emulation of synaptic plasticity, crucial for learning.
• Neuronal intrinsic plasticity, vital for learning, has been underexplored in hardware.

Purpose of the Study:

• To develop a single-cell device emulating both synaptic and intrinsic plasticity.
• To investigate the simultaneous implementation and interaction of these plasticities.
• To establish a positive feedback learning loop for enhanced neuromorphic training.

Main Methods:

• Integration of threshold switch and phase change memory into a single device (TS-PCM).
• Utilizing the threshold switch layer to emulate neuronal intrinsic plasticity (firing rate modulation).
• Employing the phase change layer for nonvolatile synaptic plasticity emulation.

Main Results:

• Demonstrated simultaneous emulation of intrinsic and synaptic plasticity in one cell.
• Established a positive feedback interaction between the two plasticity mechanisms.
• Implemented a positive feedback learning loop in a TS-PCM array for accelerated training.

Conclusions:

• The TS-PCM device successfully mimics concomitant synaptic and intrinsic plasticity.
• Simultaneous emulation creates a positive feedback loop, enhancing learning efficiency.
• This approach offers a pathway for more biologically plausible and efficient neuromorphic computing.