Linear Regression Using Quantum Annealing with Continuous Variables

TL;DR

This paper introduces a novel quantum annealing method for linear regression that directly handles continuous variables using boson systems, avoiding the need for increasing qubits to improve accuracy.

Contribution

It proposes a new QA-based linear regression approach leveraging continuous variables with boson systems, enhancing accuracy without increasing qubit count.

Findings

Achieves accurate linear regression using QA with continuous variables

Utilizes boson systems to directly manage continuous parameters

Maintains accuracy without increasing qubit numbers

Abstract

Linear regression is a data analysis technique, which is categorized as supervised learning. By utilizing known data, we can predict unknown data. Recently, researchers have explored the use of quantum annealing (QA) to perform linear regression where parameters are approximated to discrete values using binary numbers. However, this approach has a limitation: we need to increase the number of qubits to improve the accuracy. Here, we propose a novel linear regression method using QA that leverages continuous variables. In particular, the boson system facilitates the optimization of linear regression without resorting to discrete approximations, as it directly manages continuous variables while engaging in QA. The major benefit of our new approach is that it can ensure accuracy without increasing the number of qubits as long as the adiabatic condition is satisfied.