Exploiting non-linear photonic integrated circuits for reservoir computing
In a collaboration with the team of Prof. Christophe Moser from EPFL, our lithium niobate on insulator technology is used to demonstrate a reservoir computing approach (RC) exploiting the inherent optical non-linearities of the material.
The data is encoded on the spectrum of a femtosecond pulse, which is then transformed non-linearly by the photonic integrated circuit, allowing to increase the classification accuracy by 10% while reducing the number of parameters by a factor of about 20, showing that an integrated optical RC approach can attain performances on par with a digital neural network.
Check out the publication in AIP here: external page https://doi.org/10.1063/5.0158611
