TR2021-061

Inverse design for integrated photonics using deep neural network

    •  Kojima, K., Koike-Akino, T., Tang, Y., Wang, Y., "Inverse design for integrated photonics using deep neural network", Integrated Photonics Research, Silicon and Nanophotonics (IPR), DOI: 10.1364/​IPRSN.2021.IF3A.6, July 2021.
      BibTeX TR2021-061 PDF
      • @inproceedings{Kojima2021jul,
      • author = {Kojima, Keisuke and Koike-Akino, Toshiaki and Tang, Yingheng and Wang, Ye},
      • title = {{Inverse design for integrated photonics using deep neural network}},
      • booktitle = {Integrated Photonics Research, Silicon and Nanophotonics (IPR)},
      • year = 2021,
      • month = jul,
      • doi = {10.1364/IPRSN.2021.IF3A.6},
      • url = {https://www.merl.com/publications/TR2021-061}
      • }
  • MERL Contacts:
  • Research Areas:

    Artificial Intelligence, Communications, Machine Learning, Optimization, Signal Processing

Abstract:

Focusing on nanophotonic power splitters, we show that a generative neural network can design a series of devices that achieve nearly arbitrary target performance, with an excellent capability to generalize training data produced by the adjoint method.