Classification With Unimodular Matrices In Hybrid Models

Abstract

Guessing the architecture of a variational quantum circuit can be fraught with error, since determining the correct locations and types of parameterized quantum gates is often an empirical task. This work demonstrates that using a general parameterized unimodular matrix achieves a higher classification accuracy faster than comparable classical models. Variations of this ansatz and the performance results are explored and discussed to analyze this approach.