Fig. 3: The performance of full-stokes polarization and spectrum detection.

a The training and validation loss curves of the polarization detection regression task during the training. b Poincaré sphere representation of polarization states from the MLP model prediction (blue sphere) and the commercial PA. All the Stokes parameters are normalized by the optical power \({S}_{0}\) for clarity. c Predicted power \({S}_{0}\) error at different polarization states (gray dots). The intensity error of all testing samples is less than 0.026 (light blue shadow) and the average error is calculated to be 0.012. d Reconstructed spectra for a series of narrowband spectral lines across the 20 nm bandwidth. The dashed lines mark their center wavelengths of each peak. e Reconstructed spectra for continuous, broadband signals: the ASE source filtered by the PF with different wave shapes. f Reconstructed spectra for continuous, broadband signals: the ASE source filtered by the PF with different wave shapes. g The measured spectra reconstruction mean error under different noise factor σ ranging from 0.1 (-10 dB) to 0.001 (−30 dB). h Reconstructed spectra for dual spectral lines: dual peaks at 1545 and 1545.4 nm, with a spectral spacing of 400 pm.