Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

Nature Computational Science

Reducing qubit budget in quantum machine learning

Jie Pan

Published Sep 16, 2021• 0 citations
Publisher
DOI
10.1038/s43588-021-00133-9

Related papers

Advances in Quantum Machine Learning

A. S. Wilson

•Sep 1, 2021•0 citations

Toward Fair and Budget-Aware Benchmarking in Hybrid Quantum Machine Learning

Rekia Bendouma, Allal Tiberkak, Miloud Bagaa

Empirical comparisons between hybrid quan-tum-classical machine learning (HQML) models and classical baselines are often confounded by mismatched model capacity and training budgets. We propose a fair, budgetaware benchmarking protocol for HQML evaluation on classical quantum simulators. Our proposed framework ensures controlled comparisons by matching the number of trainable parameters and optimization steps under a shared early-stopping policy, while explicitly reporting the wall-clock training cost. Variational HQML models with different circuit depths are evaluated on a binary FashionMNIST task $(N=12000)$ using PCA-reduced inputs mapped to $n=8$ qubits and are compared against parameter-matched classical multilayer perceptrons.The results obtained indicate that shallow HQML models achieve statistically comparable predictive performance under strictly matched computational budgets, albeit at substantially higher computational cost. These findings emphasize the necessity of budget-aware evaluation and provide a reproducible foundation for principled HQML benchmarking.

International Conference on Pattern Analysis and Intelligent Systems (PAIS)•Apr 1, 2026•1 citation

Quantum Machine Learning on Limited-Qubit Hardware

Radosław Idzikowski, Małgorzata Jȩdruszek, Mateusz Kucharski +1 more

•Jan 1, 2026•0 citations

Improving Hardware Requirements for Fault-Tolerant Quantum Computing by Optimizing Error Budget Distributions

Tobias V. Forster, Nils Quetschlich, Mathias Soeken +1 more

Despite significant progress in quantum computing in recent years, executing quantum circuits for practical problems remains challenging due to error-prone quantum hardware. Hence, quantum error correction becomes essential but induces significant overheads in qubits and execution time, often by orders of magnitude. Obviously, these overheads must be reduced. Since many quantum applications can tolerate some noise, end users can provide a maximum tolerated error, the error budget, to be considered during compilation and execution. This error budget, or, more precisely, its distribution, can be a key factor in achieving the overhead reduction. Conceptually, an errorcorrected quantum circuit can be divided into different parts that each have a specific purpose. Errors can happen in any of these parts and their errors sum up to the mentioned error budget-but how to distribute it among them actually constitutes a degree of freedom. This work is based on the idea that some of the circuit parts can compensate for errors more efficiently than others. Consequently, these parts should contribute more to satisfy the total error budget than the parts where it is more costly. However, this poses the challenge of finding optimal distributions. We address this challenge not only by providing general guidelines on distributing the error budget, but also a method that automatically determines resource-efficient distributions for arbitrary circuits by training a machine learning model on an accumulated dataset. The approach is evaluated by analyzing the machine learning model's predictions on so far unseen data, reducing the estimated space-time costs for more than 75% of the considered quantum circuits, with an average reduction of$\mathbf{1 5. 6 \%}$, including cases without improvement, and a maximum reduction of$\mathbf{7 7. 7 \%}$.

arXiv:2509.02683•IEEE International Conference on Quantum Computing and Engineering (QCE)•Aug 30, 2025•3 citations

Disentangling encoding, preprocessing, training budget and estimator artefacts in pre-training diagnostics for hybrid quantum-classical machine learning

Ömer Doğan, İlyas Özer

A preregistered follow-up study testing six hypotheses that disentangle confounds in an earlier exploratory campaign on pre-training diagnostics for hybrid quantum-classical machine learning models. The study isolates the effects of data preprocessing, training budget, learning rate, matched circuit resources, and estimator sample size on the trainability failures previously observed at eight qubits.

•Aug 18, 2026•0 citations

Towards quantum machine learning

Michael Brooks

Nature•May 24, 2023•4 citations
Empirical comparisons between hybrid quan-tum-classical machine learning (HQML) models and classical baselines are often confounded by mismatched model capacity and training budgets. We propose a fair, budgetaware benchmarking protocol for HQML evaluation on classical quantum simulators. Our proposed framework ensures controlled comparisons by matching the number of trainable parameters and optimization steps under a shared early-stopping policy, while explicitly reporting the wall-clock training cost. Variational HQML models with different circuit depths are evaluated on a binary FashionMNIST task (N=12000)(N=12000)(N=12000) using PCA-reduced inputs mapped to n=8n=8n=8 qubits and are compared against parameter-matched classical multilayer perceptrons.The results obtained indicate that shallow HQML models achieve statistically comparable predictive performance under strictly matched computational budgets, albeit at substantially higher computational cost. These findings emphasize the necessity of budget-aware evaluation and provide a reproducible foundation for principled HQML benchmarking.
Despite significant progress in quantum computing in recent years, executing quantum circuits for practical problems remains challenging due to error-prone quantum hardware. Hence, quantum error correction becomes essential but induces significant overheads in qubits and execution time, often by orders of magnitude. Obviously, these overheads must be reduced. Since many quantum applications can tolerate some noise, end users can provide a maximum tolerated error, the error budget, to be considered during compilation and execution. This error budget, or, more precisely, its distribution, can be a key factor in achieving the overhead reduction. Conceptually, an errorcorrected quantum circuit can be divided into different parts that each have a specific purpose. Errors can happen in any of these parts and their errors sum up to the mentioned error budget-but how to distribute it among them actually constitutes a degree of freedom. This work is based on the idea that some of the circuit parts can compensate for errors more efficiently than others. Consequently, these parts should contribute more to satisfy the total error budget than the parts where it is more costly. However, this poses the challenge of finding optimal distributions. We address this challenge not only by providing general guidelines on distributing the error budget, but also a method that automatically determines resource-efficient distributions for arbitrary circuits by training a machine learning model on an accumulated dataset. The approach is evaluated by analyzing the machine learning model's predictions on so far unseen data, reducing the estimated space-time costs for more than 75% of the considered quantum circuits, with an average reduction of15.6%\mathbf{1 5. 6 \%}15.6%, including cases without improvement, and a maximum reduction of77.7%\mathbf{7 7. 7 \%}77.7%.