An electro-photo-sensitive synaptic transistor for edge neuromorphic visual systems.

Area of Science:

• Materials Science
• Neuroscience
• Electrical Engineering

Background:

• Optoelectronic artificial synapses are crucial for neuromorphic visual systems but face challenges in functionality, reliability, and mass production.
• Existing devices often lack the necessary adaptability and robustness for practical applications.

Purpose of the Study:

• To develop a highly reliable and functional electro-photo-sensitive artificial synapse.
• To explore optoelectronic synergetic modulation for reconfigurable synaptic behaviors.
• To evaluate the performance of the artificial synapse in a neuromorphic computing task.

Main Methods:

• Fabrication of an artificial synapse using amorphous Indium Gallium Zinc Oxide (InGaZnO) thin-film transistors.
• Characterization of synaptic plasticity, including excitatory and inhibitory postsynaptic currents and plasticity transitions.
• Implementation of a LeNet-5 convolutional neural network simulation for MNIST handwritten digit recognition using the artificial synapse.

Main Results:

• Demonstrated synaptic plasticity, including inhibitory/excitatory currents, frequency dependence, and plasticity transitions.
• Achieved optoelectronic synergetic modulation for reconfigurable synaptic behaviors and homeostatic regulation.
• Attained 95.99% accuracy in MNIST recognition with high tolerance to noisy inputs.

Conclusions:

• The developed InGaZnO-based artificial synapse offers enhanced functionality and reliability for neuromorphic systems.
• Optoelectronic synergetic modulation provides a novel approach for synaptic weight regulation.
• The technology shows significant commercial potential for edge neuromorphic applications.