Photonic Machine Learning with Wavelength/Frequency Multiplexing

Case ID:

This invention is an improved method of machine learning utilizing nanophotonic devices to implement neural networks. Photons from nanophotonic devices have degrees of freedom which allow for more information to be encoded in the photons. This allows the neural network to be expanded. Signals then are encoded in different wavelengths which allow for many signals to be transmitted in a single signal that must be decoded with a similar technology. This method works best for convolutional neural networks where the same set of operations is repeatedly applied to different data.



Machine learning currently is limited by the size of the neural network being worked with. This method allows for the neural network to be expanded which allows for a more efficient machine learning method. Wavelength multiplexing allows for multiple signals to be encoded in multiple wavelengths which increases the volume of data that can be processed in a time period. Machine learning is often very intense on computing and it takes a long time to build a neural network. With this invention, the computing time and power will be decreased, and it will be made more efficient and effective in the long run.



  • Machine learning
  • Computing
  • Artificial Intelligence


  • More efficient
  • Greater volume
Patent Information:
Contact For More Information:
Richard Weite
Senior Licensing Manager, College of Optical Sciences
The University of Arizona
Lead Inventor(s):
Linran Fan
Quntao Zhuang