Claudio CONTI

We discuss the links between wave propagation and spin glasses, with particular emphasis on the realization of Ising machines, i.e., optical devices to solve combinatorial optimization. Recent developments challenge the performances of quantum annealers in solving NP-complete problems, and we discuss a simple experimental approach that may potentially scale to millions of logical spins. We also review some unconventional applications of reservoir computing to study cancer morphodynamics by using simple optical neural networks with random media. Specifically, we compare the characterization of tumor spheroids obtained by conventional biophysical methods and the signals obtained by a recurrently trained disordered photonic device. We outline the relation with modern machine learning paradigms.