Large-scale photonic computing with nonlinear disordered media
Nonlinearity is essential for neural networks but extremely challenging to implement with light. Here, we demonstrate how a nonlinear disordered material can enhance the performance of state-of-the-art photonic neuromorphic computing.
This work, carried out in collaboration with Laboratoire Kastler Brossel (LKB, France), Centro Ricerche Enrico Fermi (CREF, Italy), and Tsinghua University (China), demonstrates that our nonlinear disordered media can be effectively used to implement large-scale nonlinear optical operations. The unique combination of multiple scattering and second-harmonic generation enhances the performance of various machine learning tasks.
Read more here
Find our publication here: external page https://doi.org/10.1038/s43588-024-00644-1
