Neural network for calculating direct and inverse nonlinear Fourier transform

A neural network architecture is proposed that allows a continuous nonlinear spectrum of optical signals to be predicted and an inverse nonlinear Fourier transform (NFT) to be performed for signal modulation. The average value of the relative error in predicting the continuous spectrum by the neural network when calculating the direct NFT is found to be 2.68×10−3, and the average value of the relative error in predicting the signal for the inverse NFT is 1.62×10−4.

Download paper here or here

@article{sedov2021neural,
  title={Neural network for calculating direct and inverse nonlinear Fourier transform},
  author={Sedov, Egor Valentinovich and Chekhovskoy, Igor Sergeevich and Prilepsky, Jaroslav Evgen'evich},
  journal={Quantum Electronics},
  volume={51},
  number={12},
  pages={1118},
  year={2021},
  publisher={IOP Publishing}
}