Full-stack Quantum Machine Learning in High Energy Physics

High Energy Physics is an interesting context in which to experiment with Quantum Machine Learning routines, both because we are dealing with quantum data and because the dimensionality of the problems challenges classical models. Beyond the charm of QML models, our real goal is now to understand in what concrete way these routines can be useful and competitive with respect to the classical approach. With this talk we discuss the potentialities of Qibo as full-stack environment to test and deploy QML procedures. To do this, we present a series of tests we performed on a superconducting device to improve a gradient descent training on chip with the aim of fitting the proton Parton Distribution Functions (PDFs). In particular, we focus on real-time error mitigation of both gradients and predictions estimations, proving that these techniques can be used on the current state of the qubits to perform non-trivial tasks.