DI FRANCESCO, BenedettaBenedettaDI FRANCESCODI FRISCHIA, StefanoStefanoDI FRISCHIADI RICO, GianlucaGianlucaDI RICODOLCI, MauroMauroDOLCIDI ANTONIO, IvanIvanDI ANTONIOHarris, Robert J.Robert J.HarrisTOZZI, AndreaAndreaTOZZIIUZZOLINO, MarcellaMarcellaIUZZOLINO2025-04-022025-04-02202497815106752300277-786Xhttp://hdl.handle.net/20.500.12386/37001This research explores the potential of machine learning and neural networks in recognizing the input features of aberrated wavefronts transmitted through multimode optical fibers, in view of applications for wavefront sensing in ground-based telescopes. Recent studies highlight the efficacy of multimode fibers for imaging and sensing, suggesting neural networks' effectiveness in mapping relationships between output distortions and input wavefront aberrations. The initial step of our study concerned multimode fiber propagation simulations. An input Gaussian beam was distorted with known aberrations and then sent through the fiber to analyze the effects on the output. This groundwork was used to train and validate a Convolutional Neural Network architecture. Its main role was to understand, from output images, which type of aberration was superimposed in input. We obtained promising results with test accuracy of 85% and 87%, while achieving good performance in network training and generalization.STAMPAenConference paper10.1117/12.30179442-s2.0-85205960577https://www.spiedigitallibrary.org/conference-proceedings-of-spie/13100/3017944/Machine-learning-applied-to-fiber-fed-focal-plane-wavefront-sensing/10.1117/12.3017944.shorthttps://api.elsevier.com/content/abstract/scopus_id/852059605772024SPIE13100E..6DDFIS/05 - ASTRONOMIA E ASTROFISICA