Researchers look to improve the efficiency of the artificial synapses for development of next-generation AI systems.
Researchers across the globe are trying to implement brain-inspired computing networks that are energy efficient, lightweight and adaptable as the human brain. However, mimicking the brain’s neuroplasticity is challenging in artificial synapses using ultralow energy. Neuroplasticity is the ability of the brain to change a neural network connection.
An artificial synapse consists of a gap across two neurons to allow electrical signals to pass and communicate with each other. This allows for emulation of the efficient neural signal transmission and memory formation process of the brain.
Researchers from the Singapore University of Technology and Design improved the efficiency of artificial synapses by introducing a nanoscale deposit-only-metal-electrode fabrication process for artificial synapses. Using this approach, the researchers designed a phase-change artificial synaptic device which has achieved an energy consumption of 1.8 pJ per pair-pulse-based synaptic event, which is about 82% smaller compared to traditional artificial synapses.
“The experiments have demonstrated that the artificial synapse based on phase-change materials could perform pair-pulse facilitation/depression, long-term potentiation/depression and spike timing dependent plasticity with ultralow energies. We believe our finding can provide a promising approach for developing faster, larger scale artificial synapse arrays with significantly improved performance in AI tasks.” said Desmond Loke, assistant professor at the Singapore University of Technology and Design (SUTD).
This research was published in APL Materials.