Ultrashort laser pulse characterization by neural network

The characterization of a pulse was performed by determining, using a neural network, the parameters of a function used to represent the pulse. The input to the system was a frequency resolved optical gating (FROG) trace from which features were extracted. These features formed the input to a neural network whose output was the parameters of a function which characterized the laser pulse. While such a characterization may not be an exact representation, for many real time applications sufficient accuracy is attainable. In this work, a four parameter expression was used to characterize the pulse