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

arXiv:2310.12396 · Quantum Machine Intelligence

Estimation of mutual information via quantum kernel methods

Yota Maeda, Hideaki Kawaguchi, Hiroyuki Tezuka

Published Feb 25, 2025• 6 citations
arXivHTMLPDF

Abstract

Abstract Recently, the importance of analyzing data and collecting valuable insight efficiently has been increasing in various fields. Estimating mutual information~(MI) plays a critical role in investigating the relationship among multiple random variables with a nonlinear correlation. Particularly, the task to determine whether they are independent or not is called the independence test, whose core subroutine is estimating MI from given data. It is a fundamental tool in statistics and data analysis that can be applied in a wide range of applications such as hypothesis testing, causal discovery and more. In this paper, we propose a method for estimating mutual information using the quantum kernel. We investigate the performance under various problem settings, such as different sample sizes or the shape of the probability distribution. As a result, the quantum kernel method showed higher performance than the classical one under the situation that the number of samples is small, the variance is large or the variables possess highly non-linear relationships. We discuss this behavior in terms of the central limit theorem and the structure of the corresponding quantum reproducing kernel Hilbert space.

DOI
10.1007/s42484-025-00263-7

Related papers

Classical and quantum kernel fusion for two-sample testing

Yu Terada, Yugo Ogio, Ken Arai +2 more

Two-sample tests have been extensively employed in various scientific fields and machine learning to discriminate whether two sets of samples come from the same distribution or not. Kernel-based procedures for hypothetical testing have been proposed to efficiently disentangle high-dimensional complex structures in data to obtain accurate results in a model-free way by embedding the data into the reproducing kernel Hilbert space (RKHS). While the choice of kernels plays a crucial role for their performance, little is understood about how to choose kernel especially for small datasets. Here we construct a hypothetical test which can be effective even for small datasets, based on the theoretical foundation of kernel-based tests using maximum mean discrepancy, which is called MMD-FUSE. We enhance the MMD-FUSE framework by incorporating quantum kernels and propose a novel hybrid testing strategy that fuses classical and quantum kernels. This approach creates a powerful and adaptive test by combining the domain-specific inductive biases of classical kernels with the unique expressive power of quantum kernels. We evaluate our method on various synthetic and real-world clinical datasets, and our experiments reveal two key findings: 1) With appropriate hyperparameter tuning, MMD-FUSE with quantum kernels consistently improves test power over classical counterparts, especially for small and high-dimensional data. 2) The proposed hybrid framework demonstrates remarkable robustness, adapting to different data characteristics and achieving high test power across diverse scenarios. These results highlight the potential of quantum-inspired and hybrid kernel strategies to build more effective statistical tests, offering versatile tools for data analysis where sample sizes are limited.

quant-pharXiv:2511.20941•Nov 27, 2025•0 citations

Quantum Kernel Method in the Presence of Noise

Salman Beigi

Kernel method in machine learning consists of encoding input data into a vector in a Hilbert space called the feature space and modeling the target function as a linear map on the feature space. Given a cost function, computing such an optimal linear map requires computation of a kernel matrix whose entries equal the inner products of feature vectors. In the quantum kernel method it is assumed that the feature vectors are quantum states in which case the quantum kernel matrix is given in terms of the overlap of quantum states. In practice, to estimate entries of the quantum kernel matrix one should apply, e.g., the SWAP-test and the number of such SWAP-tests is a relevant parameter in evaluating the performance of the quantum kernel method. Moreover, quantum systems are subject to noise, so the quantum states as feature vectors cannot be prepared exactly and this is another source of error in the computation of the quantum kernel matrix. Taking both the above considerations into account, we prove a bound on the performance (generalization error) of the quantum kernel method.

arXiv:2210.08476•Oct 16, 2022•1 citation

Quantum kernel method for learning graph data via quantum walk

Linyun Cao, Youle Wang

Graph kernels are essential for learning the properties of graph data, a fundamental task in the analysis of such data. However, a major limitation of current graph kernel methods is their high computational cost, which becomes prohibitive when dealing with large-scale graphs. In order to address this bottleneck, we propose a quantum algorithm designed to analyze the intrinsic structural information within graph data through an efficient quantum kernel method. The core idea of our kernel method involves encoding pertinent information from graphs into quantum states, utilizing quantum kernel functions to represent the similarities among various graphs. This process is facilitated through the application of quantum walks and quantum signal processing techniques. Building upon this framework, we propose a quantum kernel ridge regression algorithm that consists of two key steps. First, we employ our quantum kernel method to prepare a quantum state, in which the amplitudes encode the kernel functions. Second, we implement quantum feature mapping in parallel to encode the corresponding kernel matrix into the quantum circuit. Crucially, we demonstrate that the computational complexity of our regression algorithm scales primarily as $O({M}^{3/2})$ with respect to the size of the graph set $M$. This suggests that our quantum kernel method could process graph data with resource requirements scaling approximately as $O(\text{poly}log(N))$, where $N$ denotes the number of nodes in the graph. Accordingly, this underscores the advantage of our approach over classical methods in the context of large-scale graph analysis. To validate our approach, we conduct numerical experiments on chemical molecular datasets, demonstrating the method's effectiveness and advantages. In summary, this work represents a significant advancement in quantum graph learning, enhancing its potential for analyzing large-scale graph data.

Physical Review A•Sep 3, 2025•5 citations

Importance of kernel bandwidth in quantum machine learning

Ruslan Shaydulin, Stefan M. Wild

Quantum kernel methods are considered a promising avenue for applying quantum computers to machine learning problems. Identifying hyperparameters controlling the inductive bias of quantum machine learning models is expected to be crucial given the central role hyperparameters play in determining the performance of classical machine learning methods. In this work we introduce the hyperparameter controlling the bandwidth of a quantum kernel and show that it controls the expressivity of the resulting model. We use extensive numerical experiments with multiple quantum kernels and classical data sets to show consistent change in the model behavior from underfitting (bandwidth too large) to overfitting (bandwidth too small), with optimal generalization in between. We draw a connection between the bandwidth of classical and quantum kernels and show analogous behavior in both cases. Furthermore, we show that optimizing the bandwidth can help mitigate the exponential decay of kernel values with qubit count, which is the cause behind recent observations that the performance of quantum kernel methods decreases with qubit count. We reproduce these negative results and show that if the kernel bandwidth is optimized, the performance instead improves with growing qubit count and becomes competitive with the best classical methods.

arXiv:2111.05451•Physical Review A•Oct 5, 2022•100 citations

Quantum Kernels for Text Classification: A Statistical and Diagnostic Framework Revealing the Low-Data Regime

Mrugendrasinh L. Rahevar, Martin Parmar, Hemant Yadav +3 more

Quantum kernel techniques aim to leverage quantum computational capabilities on social data. However, their application to natural language processing tasks faces formidable obstacles, such as extreme dimensionality reduction (D=384→k=8), concentration of measure in quantum feature spaces, and the lack of theoretical understanding of when quantum advantages occur in kernel-based text classification. Filling this gap, we provide a comprehensive study of quantum kernels for text classification that addresses three major challenges in existing studies: general data compression approaches that ignore class structure, the lack of a predictive diagnostic toolkit, and overlooked approaches for handling concentration effects. Our main contributions include a supervised contrastive data compression approach with theoretical guarantees of kernel alignment, a five-diagnostic toolkit connecting theoretical insights with practical performance, the discovery of a small-data regime (n≤200) in which quantum methods perform comparably to classical approaches, and a transparent demonstration that quantum kernels require carefully designed settings to remain competitive with classical counterparts. Across four datasets, evaluated using five random seeds and exact paired statistical testing, we observe that quantum projected kernels achieve an accuracy of 84.5% compared to 85.1% for the classical RBF kernel (p>0.05, not statistically significant) under supervised compression settings. However, quantum methods lag behind by approximately 2%–4% at larger scales due to concentration effects, reflected in reduced off-diagonal kernel variance (0.003 vs. 0.025).

Computer Modeling in Engineering & Sciences•Jan 1, 2026•0 citations

Quantum estimation via sequential measurements

Daniel Klaus Burgarth, Vittorio Giovannetti, Airi N. Kato +1 more

The problem of estimating a parameter of a quantum system through a series of measurements performed sequentially on a quantum probe is analyzed in the general setting where the underlying statistics is explicitly non-i.i.d. We present a generalization of the central limit theorem in the present context, which under fairly general assumptions shows that as the number N of measurement data increases the probability distribution of functionals of the data (e.g., the average of the data) through which the target parameter is estimated becomes asymptotically normal and independent of the initial state of the probe. At variance with the previous studies (Guţă M 2011 Phys. Rev. A 83 062324 ; van Horssen M and Guţă M 2015 J. Math. Phys. 56 022109 ) we take a diagrammatic approach, which allows one to compute not only the leading orders in N of the moments of the average of the data but also those of the correlations among subsequent measurement outcomes. In particular our analysis points out that the latter, which are not available in usual i.i.d. data, can be exploited in order to improve the accuracy of the parameter estimation. An explicit application of our scheme is discussed by studying how the temperature of a thermal reservoir can be estimated via sequential measurements on a quantum probe in contact with the reservoir.

arXiv:1507.07634•New Journal of Physics•Nov 27, 2015•38 citations
Abstract Recently, the importance of analyzing data and collecting valuable insight efficiently has been increasing in various fields. Estimating mutual information~(MI) plays a critical role in investigating the relationship among multiple random variables with a nonlinear correlation. Particularly, the task to determine whether they are independent or not is called the independence test, whose core subroutine is estimating MI from given data. It is a fundamental tool in statistics and data analysis that can be applied in a wide range of applications such as hypothesis testing, causal discovery and more. In this paper, we propose a method for estimating mutual information using the quantum kernel. We investigate the performance under various problem settings, such as different sample sizes or the shape of the probability distribution. As a result, the quantum kernel method showed higher performance than the classical one under the situation that the number of samples is small, the variance is large or the variables possess highly non-linear relationships. We discuss this behavior in terms of the central limit theorem and the structure of the corresponding quantum reproducing kernel Hilbert space.
Two-sample tests have been extensively employed in various scientific fields and machine learning to discriminate whether two sets of samples come from the same distribution or not. Kernel-based procedures for hypothetical testing have been proposed to efficiently disentangle high-dimensional complex structures in data to obtain accurate results in a model-free way by embedding the data into the reproducing kernel Hilbert space (RKHS). While the choice of kernels plays a crucial role for their performance, little is understood about how to choose kernel especially for small datasets. Here we construct a hypothetical test which can be effective even for small datasets, based on the theoretical foundation of kernel-based tests using maximum mean discrepancy, which is called MMD-FUSE. We enhance the MMD-FUSE framework by incorporating quantum kernels and propose a novel hybrid testing strategy that fuses classical and quantum kernels. This approach creates a powerful and adaptive test by combining the domain-specific inductive biases of classical kernels with the unique expressive power of quantum kernels. We evaluate our method on various synthetic and real-world clinical datasets, and our experiments reveal two key findings: 1) With appropriate hyperparameter tuning, MMD-FUSE with quantum kernels consistently improves test power over classical counterparts, especially for small and high-dimensional data. 2) The proposed hybrid framework demonstrates remarkable robustness, adapting to different data characteristics and achieving high test power across diverse scenarios. These results highlight the potential of quantum-inspired and hybrid kernel strategies to build more effective statistical tests, offering versatile tools for data analysis where sample sizes are limited.
Kernel method in machine learning consists of encoding input data into a vector in a Hilbert space called the feature space and modeling the target function as a linear map on the feature space. Given a cost function, computing such an optimal linear map requires computation of a kernel matrix whose entries equal the inner products of feature vectors. In the quantum kernel method it is assumed that the feature vectors are quantum states in which case the quantum kernel matrix is given in terms of the overlap of quantum states. In practice, to estimate entries of the quantum kernel matrix one should apply, e.g., the SWAP-test and the number of such SWAP-tests is a relevant parameter in evaluating the performance of the quantum kernel method. Moreover, quantum systems are subject to noise, so the quantum states as feature vectors cannot be prepared exactly and this is another source of error in the computation of the quantum kernel matrix. Taking both the above considerations into account, we prove a bound on the performance (generalization error) of the quantum kernel method.
Graph kernels are essential for learning the properties of graph data, a fundamental task in the analysis of such data. However, a major limitation of current graph kernel methods is their high computational cost, which becomes prohibitive when dealing with large-scale graphs. In order to address this bottleneck, we propose a quantum algorithm designed to analyze the intrinsic structural information within graph data through an efficient quantum kernel method. The core idea of our kernel method involves encoding pertinent information from graphs into quantum states, utilizing quantum kernel functions to represent the similarities among various graphs. This process is facilitated through the application of quantum walks and quantum signal processing techniques. Building upon this framework, we propose a quantum kernel ridge regression algorithm that consists of two key steps. First, we employ our quantum kernel method to prepare a quantum state, in which the amplitudes encode the kernel functions. Second, we implement quantum feature mapping in parallel to encode the corresponding kernel matrix into the quantum circuit. Crucially, we demonstrate that the computational complexity of our regression algorithm scales primarily as O(M3/2)O({M}^{3/2})O(M3/2) with respect to the size of the graph set MMM. This suggests that our quantum kernel method could process graph data with resource requirements scaling approximately as O(polylog(N))O(\text{poly}log(N))O(polylog(N)), where NNN denotes the number of nodes in the graph. Accordingly, this underscores the advantage of our approach over classical methods in the context of large-scale graph analysis. To validate our approach, we conduct numerical experiments on chemical molecular datasets, demonstrating the method's effectiveness and advantages. In summary, this work represents a significant advancement in quantum graph learning, enhancing its potential for analyzing large-scale graph data.
Quantum kernel methods are considered a promising avenue for applying quantum computers to machine learning problems. Identifying hyperparameters controlling the inductive bias of quantum machine learning models is expected to be crucial given the central role hyperparameters play in determining the performance of classical machine learning methods. In this work we introduce the hyperparameter controlling the bandwidth of a quantum kernel and show that it controls the expressivity of the resulting model. We use extensive numerical experiments with multiple quantum kernels and classical data sets to show consistent change in the model behavior from underfitting (bandwidth too large) to overfitting (bandwidth too small), with optimal generalization in between. We draw a connection between the bandwidth of classical and quantum kernels and show analogous behavior in both cases. Furthermore, we show that optimizing the bandwidth can help mitigate the exponential decay of kernel values with qubit count, which is the cause behind recent observations that the performance of quantum kernel methods decreases with qubit count. We reproduce these negative results and show that if the kernel bandwidth is optimized, the performance instead improves with growing qubit count and becomes competitive with the best classical methods.
Quantum kernel techniques aim to leverage quantum computational capabilities on social data. However, their application to natural language processing tasks faces formidable obstacles, such as extreme dimensionality reduction (D=384→k=8), concentration of measure in quantum feature spaces, and the lack of theoretical understanding of when quantum advantages occur in kernel-based text classification. Filling this gap, we provide a comprehensive study of quantum kernels for text classification that addresses three major challenges in existing studies: general data compression approaches that ignore class structure, the lack of a predictive diagnostic toolkit, and overlooked approaches for handling concentration effects. Our main contributions include a supervised contrastive data compression approach with theoretical guarantees of kernel alignment, a five-diagnostic toolkit connecting theoretical insights with practical performance, the discovery of a small-data regime (n≤200) in which quantum methods perform comparably to classical approaches, and a transparent demonstration that quantum kernels require carefully designed settings to remain competitive with classical counterparts. Across four datasets, evaluated using five random seeds and exact paired statistical testing, we observe that quantum projected kernels achieve an accuracy of 84.5% compared to 85.1% for the classical RBF kernel (p>0.05, not statistically significant) under supervised compression settings. However, quantum methods lag behind by approximately 2%–4% at larger scales due to concentration effects, reflected in reduced off-diagonal kernel variance (0.003 vs. 0.025).
The problem of estimating a parameter of a quantum system through a series of measurements performed sequentially on a quantum probe is analyzed in the general setting where the underlying statistics is explicitly non-i.i.d. We present a generalization of the central limit theorem in the present context, which under fairly general assumptions shows that as the number N of measurement data increases the probability distribution of functionals of the data (e.g., the average of the data) through which the target parameter is estimated becomes asymptotically normal and independent of the initial state of the probe. At variance with the previous studies (Guţă M 2011 Phys. Rev. A 83 062324 ; van Horssen M and Guţă M 2015 J. Math. Phys. 56 022109 ) we take a diagrammatic approach, which allows one to compute not only the leading orders in N of the moments of the average of the data but also those of the correlations among subsequent measurement outcomes. In particular our analysis points out that the latter, which are not available in usual i.i.d. data, can be exploited in order to improve the accuracy of the parameter estimation. An explicit application of our scheme is discussed by studying how the temperature of a thermal reservoir can be estimated via sequential measurements on a quantum probe in contact with the reservoir.