Exponential separations between classical and quantum learners

TL;DR

This paper explores the potential for quantum learning algorithms to achieve exponential speedups over classical algorithms in identifying data-generating functions, especially in natural scientific settings, by analyzing theoretical foundations and proposing new separations.

Contribution

It introduces two new learning separations where classical difficulty is mainly in identifying the data-generating function, and discusses leveraging quantum hardness assumptions for natural data scenarios.

Findings

Quantum algorithms can outperform classical ones in identifying data-generating functions.

Existing quantum speedups often rely on classical hardness of function evaluation, not identification.

New separations show quantum advantage when classical difficulty centers on function identification.

Abstract

Despite significant effort, the quantum machine learning community has only demonstrated quantum learning advantages for artificial cryptography-inspired datasets when dealing with classical data. In this paper we address the challenge of finding learning problems where quantum learning algorithms can achieve a provable exponential speedup over classical learning algorithms. We reflect on computational learning theory concepts related to this question and discuss how subtle differences in definitions can result in significantly different requirements and tasks for the learner to meet and solve. We examine existing learning problems with provable quantum speedups and find that they largely rely on the classical hardness of evaluating the function that generates the data, rather than identifying it. To address this, we present two new learning separations where the classical difficulty primarily lies in identifying the function generating the data. Furthermore, we explore computational hardness assumptions that can be leveraged to prove quantum speedups in scenarios where data is quantum-generated, which implies likely quantum advantages in a plethora of more natural settings (e.g., in condensed matter and high energy physics). We also discuss the limitations of the classical shadow paradigm in the context of learning separations, and how physically-motivated settings such as characterizing phases of matter and Hamiltonian learning fit in the computational learning framework.