Title: Machine learning analysis of rogue solitons in supercontinuum generation

Abstract: Supercontinuum generation is a highly nonlinear process that exhibits unstable and chaotic characteristics when developing from long pump pulses injected into the anomalous dispersion regime of an optical fiber. A particular feature associated with this regime is the long-tailed "rogue wave"-like statistics of the spectral intensity on the long wavelength edge of the supercontinuum, linked to the generation of a small number of "rogue solitons" with extreme red-shifts. Here, we apply machine learning to analyze the characteristics of these solitons at the edge of the supercontinuum spectrum, and show how supervised learning can train a neural network to predict the peak power, duration, and temporal delay of these solitons from only the supercontinuum spectral intensity without phase information. The network accurately predicts soliton characteristics for a wide range of scenarios, from the onset of spectral broadening dominated by pure modulation instability to near octave-spanning supercontinuum with distinct rogue solitons.