SciGroveBeta
Quantum

Evidence of Quantum Machine Learning Advantage with Tens of Noisy Qubits

Onur Danaci, Yash J. Patel, Riccardo Molteni, Evert van Nieuwenburg, Vedran Dunjko, Jan A. Krzywda

Featured May 24, 2026

Original Paper

This analysis was generated by SciGrove. Upload your own PDFs or enter a DOI — and get the same AI breakdown on any paper.

Get started

AI-generated analysis — This is SciGrove's AI interpretation of the paper, not peer-reviewed content. Always refer to the original paper.

Simply

Even with today's imperfect quantum computers, a new study shows that using quantum processing directly on special quantum data can solve problems much faster than turning that data into regular computer information first, especially when certain types of noise are present.

In depth
The paper provides empirical evidence that quantum machine learning can outperform classical methods even with noisy quantum devices at a practical scale of 30-40 qubits. It shows that a fully quantum protocol (FQ), which processes quantum data coherently, requires significantly fewer measurements than optimized measure-first protocols (MF) that convert quantum data to classical data before processing. This advantage is particularly pronounced under thermal relaxation noise, where FQ remains robust while MF methods struggle.

Key Takeaways

  • 1
    A clear performance separation is demonstrated between coherent quantum processing and classical fixed-measurement schemes at a practical scale of 30-40 noisy qubits.
  • 2
    The quantum advantage actively widens under specific physical noise, such as thermal relaxation, where the fully quantum protocol remains robust while classical methods severely degrade.
  • 3
    Matching the performance of the noisy fully quantum protocol requires years of data acquisition for optimized classical measure-first strategies, highlighting a shift in the fundamental bottleneck to data acquisition.

Conceptual Flow

HIGH LEVEL
1
Methodology: Comparing Quantum vs. Classical Learning

The study compares two ways to learn from quantum data: one uses a quantum computer to process the data directly, and the other measures the quantum data first to turn it into regular computer data before processing.

Quantum Data
Compare Two Ways
Quantum Processing
Measure & Classic Processing
2
Results: Quantum Advantage in Data Acquisition

They found that the quantum processing method needs far fewer tries (measurements) to learn correctly than the measure-first method, especially as the problem gets bigger and noisier.

Problem Size
Noise Level
Quantum vs Classic Effort
Quantum Needs Less
Classic Needs More