Quantum Machine Learning With Canonical Variables

TL;DR

This paper proposes a quantum machine learning platform based on electromagnetic field control over charged particles, operating on observables with exactly solvable models for tasks like regression and classification.

Contribution

It introduces a novel quantum machine learning approach using canonical variables and electromagnetic fields, with models that are exactly solvable and applicable to practical learning tasks.

Findings

Models are exactly solvable, providing clear solutions.

The platform can perform regression and classification tasks.

Physical realization is feasible in ion traps or particle confinement devices.

Abstract

Utilising dynamic electromagnetic field control over charged particles serves as the basis for a quantum machine learning platform that operates on observables rather than directly on states. Such a platform can be physically realised in ion traps or particle confinement devices that utilise electromagnetic fields as the source of control. The electromagnetic field acts as the ansatz within the learning algorithm. The models discussed are exactly solvable, with exact solutions serving as precursors for learning tasks to emerge, including regression and classification algorithms as particular cases. This approach is considered in terms of canonical variables with semi-classical behaviour, disregarding relativistic degrees of freedom.