TR2018-183

Deep Neural Network Inverse Modeling for Integrated Photonics


Abstract:

We propose a deep neural network model that instantaneously predicts the optical response of nanopatterned silicon photonic power splitter topologies, and inversely approximates compact (2.6 x 2.6 um2) and efficient (above 92%) power splitters for target splitting ratios.

 

  • Related News & Events

    •  NEWS    MERL presenting 4 papers at OFC, including an invited talk
      Date: March 3, 2019 - March 7, 2019
      Where: San Diego, CA
      MERL Contacts: Toshiaki Koike-Akino; Chungwei Lin; Kieran Parsons; Bingnan Wang; Ye Wang
      Research Areas: Communications, Machine Learning, Optimization, Signal Processing
      Brief
      • MERL researchers are presenting 4 papers at the OSA Optical Fiber Conference (OFC), which is being held in San Diego from March 3-7, 2019. Topics to be presented include recent advances in nonbinary polar codes, joint polar-coded shaping, and deep learning-based photonics circuit design. Additionally, recent work on multiset-partition distribution matching is presented as an invited talk.

        OFC is the flagship conference of the OSA, and the world's most comprehensive technical conference focused on the research advances and latest technological development in optics and photonics. The event attracts more than 10000 participants each year.
    •