Problem-Dependent Power of Quantum Neural Networks on Multi-Class Classification

TL;DR

This paper systematically analyzes the problem-dependent capabilities of quantum neural classifiers in multi-class tasks, revealing key factors like training loss dominance and risk behaviors, and proposes a method to evaluate their effectiveness over classical models.

Contribution

It introduces a risk-based analysis of QNNs, uncovers their unique risk curve behavior, and provides a practical approach to assess their potential advantages over classical classifiers.

Findings

QCs' power is mainly influenced by training loss.

QCs exhibit a U-shaped risk curve, unlike classical models.

The proposed method effectively predicts when QCs outperform classical classifiers.

Abstract

Quantum neural networks (QNNs) have become an important tool for understanding the physical world, but their advantages and limitations are not fully understood. Some QNNs with specific encoding methods can be efficiently simulated by classical surrogates, while others with quantum memory may perform better than classical classifiers. Here we systematically investigate the problem-dependent power of quantum neural classifiers (QCs) on multi-class classification tasks. Through the analysis of expected risk, a measure that weighs the training loss and the generalization error of a classifier jointly, we identify two key findings: first, the training loss dominates the power rather than the generalization ability; second, QCs undergo a U-shaped risk curve, in contrast to the double-descent risk curve of deep neural classifiers. We also reveal the intrinsic connection between optimal QCs and the Helstrom bound and the equiangular tight frame. Using these findings, we propose a method that uses loss dynamics to probe whether a QC may be more effective than a classical classifier on a particular learning task. Numerical results demonstrate the effectiveness of our approach to explain the superiority of QCs over multilayer Perceptron on parity datasets and their limitations over convolutional neural networks on image datasets. Our work sheds light on the problem-dependent power of QNNs and offers a practical tool for evaluating their potential merit.