Long-Term Homeostatic Properties Complementary to Hebbian Rules in CuPc-Based Multifunctional Memristor

Area of Science:

• Materials Science
• Neuroscience
• Computer Engineering

Background:

• Current memristor simulations for artificial synapses primarily use basic Hebbian learning rules.
• These simplified rules neglect long-term homeostasis, potentially causing neural activity collapse in realistic networks.

Purpose of the Study:

• To develop organic memristors capable of implementing both Hebbian rules and homeostatic plasticity.
• To create artificial synapses that support long-term neural network stability and comprehensive functions.

Main Methods:

• Fabrication of organic memristors using copper phthalocyanine (CuPc).
• Implementation of excitatory and inhibitory conductivity to model Hebbian and homeostatic plasticity.
• Testing adaptive homeostasis in thicker memristor samples under varying stimuli.
• Incorporating bio-inspired habituation and sensitization functions.

Main Results:

• The developed CuPc-based memristors successfully implement both Hebbian and homeostatic plasticity.
• Demonstrated adaptive homeostasis where neural activity adjusts to stimuli and recovers.
• Bio-inspired habituation and sensitization functions outperformed conventional algorithms.
• Mutual regulation between plasticity mechanisms achieved homeostasis.

Conclusions:

• A novel versatile memristor with advanced synaptic homeostasis has been developed.
• This technology offers a more robust framework for artificial neural networks.
• The memristors enable comprehensive neural functions beyond simplified models.