Abstract
We demonstrate that optical time-dynamics are equivalent to a recurrent neural network and that they can be trained for high-performance on complex classification tasks, paving the way for passive analog machine learning processors.
© 2020 The Author(s)
PDF Article | Presentation VideoMore Like This
Stephan Pachnicke and Shi Li
T1B.3 Asia Communications and Photonics Conference (ACPC) 2020
Christine Tremblay
C1F_2 Conference on Lasers and Electro-Optics/Pacific Rim (CLEOPR) 2020
Hyung Joon Cho, Siddharth Varughese, Daniel Lippiatt, and Stephen E. Ralph
M2J.5 Optical Fiber Communication Conference (OFC) 2020