SAQNN: Spectral Adaptive Quantum Neural Network as a Universal Approximator

TL;DR

This paper introduces SAQNN, a quantum neural network model with universal approximation capabilities, adaptable to various functions, and more efficient than classical neural networks in certain approximation tasks.

Contribution

The paper presents a constructive quantum neural network model that proves the universal approximation property and demonstrates asymptotic advantages over classical neural networks.

Findings

Proves SAQNN can approximate any square-integrable function.

Supports switching function bases for adaptability.

Achieves optimal parameter complexity for Sobolev functions.

Abstract

Quantum machine learning (QML), as an interdisciplinary field bridging quantum computing and machine learning, has garnered significant attention in recent years. Currently, the field as a whole faces challenges due to incomplete theoretical foundations for the expressivity of quantum neural networks (QNNs). In this paper we propose a constructive QNN model and demonstrate that it possesses the universal approximation property (UAP), which means it can approximate any square-integrable function up to arbitrary accuracy. Furthermore, it supports switching function bases, thus adaptable to various scenarios in numerical approximation and machine learning. Our model has asymptotic advantages over the best classical feed-forward neural networks in terms of circuit size and achieves optimal parameter complexity when approximating Sobolev functions under $L_2$ norm.