This project will explore the utilization of deep learning techniques, including Deep Neural Networks (DNN), Convolutional Neural Networks (CNN), and Recurrent Neural Networks – Long Short-Term Memory (RNN-LSTM), for characterizing ultrafast optical pulses using nonlinear photonic crystal fibers (NL-PCF). By employing a femtosecond laser system and a commercial NL-PCF, we will investigate how variations in pulse widths and nonlinear length (L_NL) will influence spectral and supercontinuum (SC) evolution characteristics. Our experimental setup, integrated with a fully automated control system, will generate a comprehensive dataset of final spectra and SC evolution images across various pulse widths and L_NL values. Additionally, we will employ the RNN-LSTM algorithm to augment our dataset for training the DNN and CNN models, thereby improving their predictive capabilities regarding laser pulse characteristics and coherence loss effects. Our expected results will highlight the superiority of the DNN model in predicting pulse widths from SC data under coherent conditions, while the CNN will excel in scenarios with degraded coherence. This study will not only advance our understanding of SC generation and pulse characterization but will also underscore the potential of deep learning in the analysis of nonlinear optical phenomena.