Fully forward mode training for optical neural networks

Abstract Optical computing promises to improve the speed and energy efficiency of machine learning applications 1–6 . However, current approaches to efficiently train these models are limited by in silico emulation on digital computers. Here we develop a method called fully forward mode (FFM) learning, which implements the compute-intensive training process on the physical system. The majority of the machine learning operations are thus efficiently conducted in parallel on site, alleviating numerical modelling constraints. In free-space and integrated photonics, we experimentally demonstrate optical systems with state-of-the-art performances for a given network size. FFM learning shows training the deepest…

• CA
  China Association for Science and Technology

  Award: 2023QNRC001
• NN
  National Natural Science Foundation of China

  Awards: 2023QNRC001, 62088102, 62125106
• CP
  China Postdoctoral Science Foundation

  Award: GZB20230372
• NS
  National Science and Technology Major Project

  Award: 2021ZD0109901