{"id":"arxiv/2604.12441","url":"https://arxiv.org/abs/2604.12441","tags":["quant-ph"],"score":0,"title":"Efficient classical training of model-free quantum photonic reservoir","domain":"quant-ph","source":"arxiv","authors":["Rosario Di Bartolo","Valeria Cimini","Giorgio Minati","Danilo Zia","Luca Innocenti","Salvatore Lorenzo","Gabriele Lo Monaco","Nicolò Spagnolo","Taira Giordani","G. Massimo Palma","Mauro Paternostro","Alessandro Ferraro","Fabio Sciarrino"],"reviews":0,"subject":"quant-ph","summary":"Model-independent estimation of the properties of quantum states is a central challenge in quantum technologies, as experimental imperfections, drifts, and imprecise models of the actual quantum dynamics inevitably hinder accurate reconstructions. Here, we introduce a training strategy for photonic quantum extreme learning machines in which both the learning stage and the optimization of the measurement settings are performed entirely with classical light, while inference is carried out on genuinely quantum states. The protocol is based on the identity between the normalized output…","category":"research","comments":0,"disputed":false,"division":"physics","mentions":0,"discussed":0,"published":"2026-04-14","reactions":{}}