Universal Photonic Neural Networks with Quantum-Free Data Reuploading

    •  Kojima, K., Koike-Akino, T., "Universal Photonic Neural Networks with Quantum-Free Data Reuploading", SPIE Photonics for Quantum, June 2024.
      BibTeX TR2024-074 PDF
      • @inproceedings{Kojima2024jun,
      • author = {Kojima, Keisuke and Koike-Akino, Toshiaki}},
      • title = {Universal Photonic Neural Networks with Quantum-Free Data Reuploading},
      • booktitle = {SPIE Photonics for Quantum},
      • year = 2024,
      • month = jun,
      • url = {}
      • }
  • MERL Contact:
  • Research Area:

    Electronic and Photonic Devices


The data reuploading trick was originally proposed for quantum computing to achieve the universal approximation property. In this paper, we introduce data reuploading to realize universal non-quantum photonic computing with practical photonic integrated circuits (PICs). We aim to comprehensively discuss the various advantages and implementation considerations of this approach. Our framework can eliminate the need of quantum squeezed lights, photon counters, and nonlinear photonics, which have been essential for enabling photonic neural networks in conventional configurations. Additionally, we explore ways to minimize the optical components by combining multiple functionalities into a single phase shifter, showing competitive performance when compared to using the same number of phase shifters, all without employing any nonlinear photonic devices. Considering these char- acteristics, our investigation into the use of PICs for data reuploading presents a novel architectural approach to realize photonic neural networks. This approach embodies unique features that distinctly set it apart from traditional photonic neural networks.