Quantum Adiabatic Feature Selection

TL;DR

This paper proposes a quantum adiabatic algorithm for feature selection in machine learning, offering a potentially faster solution to the NP-hard problem by leveraging quantum computation to improve efficiency.

Contribution

It introduces a novel quantum adiabatic algorithm for bi-quadratic feature selection, demonstrating improved time complexity over classical methods.

Findings

Quantum algorithm achieves $O(1/g_{min}^{2})$ time complexity.

Potential for faster feature selection in high-dimensional data.

Addresses NP-hardness of feature selection with quantum approach.

Abstract

Dimensionality reduction is the fundamental problem for machine learning and pattern recognition. During data preprocessing, the feature selection is often demanded to reduce the computational complexity. The problem of feature selection is categorized as a NP optimization problem. Exhaustive search of huge set of features takes huge amount of time on classical computer. In the present paper we discuss the role of quantum adiabatic computation to perform feature selection with bi-quadratic optimization and provide a quantum feature selection algorithm. Our algorithm runs with the quantum adiabatic time complexity bound $O(1/g_{min}^{2})$, which is better than classical approach for bi-quadratic feature selection.