Workshops Archive

Gradient-Free Optimization of Quantum Circuits Parameters for Hybrid Machine Learning


Workshop: 17th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Heterogeneous Systems (ScalAH'26)

Authors: Maxence Vandromme (RIKEN) and Miwako Tsuji (University of Tsukuba, RIKEN)

Abstract: Quantum Machine Learning generally aims to build more accurate models by using quantum mechanical principles to better capture complex correlation across data features. Existing methods scale poorly with the number of both data samples and features, making them very challenging to train on real quantum devices beyond trivial datasets. We introduce a new hybrid quantum-classical algorithm for supervised classification, which uses gradient-free optimization to learn the quantum circuit parameters and builds the final classical model from the quantum circuits measurements. We experiment over a variety of datasets and show that this algorithm performs better than standard Quantum ML algorithms in most cases, while being more scalable to larger datasets and fully trainable on current quantum devices.


Back to 17th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Heterogeneous Systems (ScalAH'26) Archive Listing

Back to Full Workshop Archive Listing