Open AccessPhotonic pattern reconstruction enabled by on-chip online learning and inference
Author(s) -
Bicky A. Márquez,
Zhimu Guo,
Hugh Morison,
Sudip Shekhar,
Lukas Chrostowski,
Paul R. Prucnal,
Bhavin J. Shastri
Publication year - 2021
Publication title -
jphys photonics
Language(s) - English
Resource type - Journals
ISSN - 2515-7647
DOI - 10.1088/2515-7647/abe3d9
Subject(s) - neuromorphic engineering , photonics , silicon photonics , computer science , artificial neural network , exploit , inference , bandwidth (computing) , hopfield network , chip , electronic engineering , latency (audio) , computer architecture , artificial intelligence , telecommunications , engineering , optoelectronics , physics , computer security
Recent investigations in neuromorphic photonics exploit optical device physics for neuron models, and optical interconnects for distributed, parallel, and analog processing. Integrated solutions enabled by silicon photonics enable high-bandwidth, low-latency and low switching energy, making it a promising candidate for special-purpose artificial intelligence hardware accelerators. Here, we experimentally demonstrate a silicon photonic chip that can perform training and testing of a Hopfield network, i.e. recurrent neural network, via vector dot products. We demonstrate that after online training, our trained Hopfield network can successfully reconstruct corrupted input patterns.