Fig. 1. Experimental setup of the TPA-based (a) intensity and (b) interferometric autocorrelator using a GaAs photodiode.

Fig. 2. Example of a simple typical NN. (a) The basic process of machine learning and (b) the brief structure of the NN.

Fig. 3. Interferometric autocorrelation signals measured by the (a) TPA photodiode-based autocorrelator and (b) SHG autocorrelator.

Fig. 4. TPA-based autocorrelation traces curve-fitted by the NN with the performance of (a) 10−5, (b) 10−6, (c) 10−7, and (d) 10−8.

Fig. 5. Training process and error analysis. (a) The gradient at each epoch, (b) MSE at each epoch, (c) comparison between the output and target, and (d) structure of NN for this measurement system.

Fig. 6. Gaussian-fitted autocorrelation traces by the L-M method. (a) Bad curve-fitting result for a signal with high noise and (b) good curve-fitting result for a signal with lower noise.