Nonlinear optical system and method for optical information processing

An optical information processing method, comprising processing encoded information input in form of pulsed light in a nonlinear element, and reading out for spectro-temporal feature extraction, comprising training to a user-defined task by at least one of:

• i) recording an output and using machine-learning to retrieve an input-specific response, in offline-training configuration;
• ii) using a tunable temporal pulse splitter between the input and the nonlinear element, evaluating an output and feeding back to the tunable temporal splitter for improving amplitude or phase features distinguishability, in offline-training;
• iii) using a tunable spectral and/or temporal filter after the nonlinear element to extract the amplitude or phase features in online output training.

An optical information processing method, comprising processing information input in form of pulsed light in a nonlinear element, and reading out for spectro-temporal feature extraction, the method comprising training to a user-defined task by at least one of:

• i) recording an output and using machine-learning to retrieve an input-specific response;
• ii) using a tunable spectral and/or temporal filter or temporal pulse splitter between the input and the nonlinear element, evaluating an output and feeding back to the tunable temporal splitter for improving feature amplitude or phase features distinguishability;
• iii) using a tunable spectral and/or temporal filter after the nonlinear element to extract the amplitude or phase features.