Related Experiment Video
Updated: Sep 12, 2025
08:58
Silicon Nanowires and Optical Stimulation for Investigations of Intra- and Intercellular Electrical Coupling
Published on: January 28, 2021
4.6K
Neural network connectivity by optical broadcasting between III-V nanowires.
Kristians Draguns1, Vidar Flodgren2, David Winge2
1University of Latvia, Riga, Latvia.
Nanophotonics (Berlin, Germany)
|August 7, 2025
Summary
This study introduces artificial neurons using semiconductor nanowires that communicate via light broadcasting, eliminating physical wiring. This novel approach enables complex connectivity for advanced artificial neural network designs.
Area of Science:
- Nanophotonics
- Artificial Intelligence
- Materials Science
Background:
- Biological neural networks rely on numerous connections, difficult to replicate artificially.
- Traditional artificial neural networks use physical wiring, limiting scalability and complexity.
Purpose of the Study:
- To explore artificial neurons using light-emitting/receiving nanowires for signal broadcasting.
- To demonstrate tunable connection weights through geometric light patterns.
- To simulate a reservoir neural network for chaotic time series prediction.
Main Methods:
- Utilized III-V semiconductor nanowires in a quasi-2D waveguide for light-based communication.
- Simulated anisotropic light emission and wavelength-specific absorption.
- Determined connection strength via nanowire rotation and separation.
- Modeled a reservoir neural network with hexagonal nanowire patterns.
Main Results:
- Achieved complex and variable connection weight distributions by tailoring nanowire geometry.
- Demonstrated that wavelength matching is crucial for network design.
- Successfully simulated a reservoir neural network capable of chaotic time series prediction.
Conclusions:
- Nanowire-based optical communication offers a viable solution for complex artificial neural networks.
- Geometric control of light patterns allows for highly adaptable neural connectivity.
- The proposed design is compatible with silicon substrates and nanophotonic integration.
