Evolving ultrafast laser information by a learning genetic algorithm combined with a knowledge base

A genetic algorithm (GA) with learning ability has been developed to retrieve the simulated ultrafast laser traces from a second-harmonic generation frequency-resolved optical gating measurement. This learning system handles the trace feature representation, storage, and utilization for helping GA evolution. By properly storing the features in a previously established knowledge base, the system can reuse previous experiences in every new evolution. Its time cost for the same error order is proved to be lower than that for the same GA without learning ability.