Tegin, UgurYildirim, MustafaOguz, IlkerMoser, ChristophePsaltis, Demetri2022-09-122022-09-122022-09-122021-01-0110.1364/CLEO_SI.2021.SW3R.3https://infoscience.epfl.ch/handle/20.500.14299/190604WOS:000831479800395A novel optical computing framework by harnessing spatiotemporal nonlinear effects of multimode fibers for machine learning is presented. With linear and nonlinear interactions of the spatial fiber modes, a brain-inspired computation engine is experimentally realized.Engineering, Electrical & ElectronicOpticsPhysics, AppliedEngineeringOpticsPhysicsMachine learning with multimode fiberstext::conference output::conference proceedings::conference paper