Ryan Hamerly, Alex Sludds, Liane Bernstein, Marin Soljačić, and Dirk Englund

DOI:https://doi.org/10.1103/PhysRevX.9.021032

Abstract:

We present a coherent optical neural-network accelerator based on homodyne detection that is scalable to large (N > 1000) networks, and analyze the fundamental quantum limits to its energy efficiency. OCIS codes: 200.4260, 200.4860, 120.2920