Abstract The importance of analyzing nontrivial datasets when testing quantum machine learning (QML) models is becoming increasingly prominent in literature, yet a cohesive framework for understanding dataset characteristics remains elusive. In this work, we introduce a data-generation tool that allows to construct semi-artificial classical datasets tailored to quantum kernel methods (QKMs). Using this tool, we show that on fully classical datasets, QKMs can require fewer training examples than classical kernels to reach comparable error, providing clear empirical evidence that data-efficient learning with quantum models is possible on classical data. The main motivation behind this tool is to enable the community to perform controlled studies to figure out which dataset characteristics are particularly fitting for quantum models by tuning the data-generation procedure. Additionally, our study brings a spectral-bias–based generalization metric from classical kernel methods into the QML domain and shows that the performance predicted by this metric aligns closely with empirical results, thereby closing an important gap between theory and practice in QML generalization. Our tool paves the way for a systematic exploration of dataset complexities. This could potentially contribute to a deeper understanding of the generalization benefits of QKM models (extendable to a broader family of QML models) and shifts the search for quantum advantage from ad hoc benchmark hunting to principled dataset design.
Abstract The importance of analyzing nontrivial datasets when testing quantum machine learning (QML) models is becoming increasingly prominent in literature, yet a cohesive framework for understanding dataset characteristics remains elusive. In this work, we introduce a data-generation tool that allows to construct semi-artificial classical datasets tailored to quantum kernel methods (QKMs). Using this tool, we show that on fully classical datasets, QKMs can require fewer training examples than classical kernels to reach comparable error, providing clear empirical evidence that data-efficient learning with quantum models is possible on classical data. The main motivation behind this tool is to enable the community to perform controlled studies to figure out which dataset characteristics are particularly fitting for quantum models by tuning the data-generation procedure. Additionally, our study brings a spectral-bias–based generalization metric from classical kernel methods into the QML domain and shows that the performance predicted by this metric aligns closely with empirical results, thereby closing an important gap between theory and practice in QML generalization. Our tool paves the way for a systematic exploration of dataset complexities. This could potentially contribute to a deeper understanding of the generalization benefits of QKM models (extendable to a broader family of QML models) and shifts the search for quantum advantage from ad hoc benchmark hunting to principled dataset design.
Abstract Quantum Machine Learning (QML) holds the promise of improving conventional machine learning, but the conditions under which an advantage can be obtained are still unclear. Although many studies have been conducted in the literature, there is no conclusive evidence that allows us to determine which types of classical datasets or which problem complexities benefit from the use of quantum techniques. Moreover, there have been contradicting findings when dealing with small or unbalanced datasets. In order to systematically approach this challenge and clarify the need for quantum properties in specific tasks, we propose the adoption of a framework capable of exploring the solution space of quantum feature maps and providing solid evidence of their potential advantages. Our framework comprises three main components: (a) the construction of datasets designed to highlight the specific characteristics that are expected to be relevant in quantum machine learning tasks; (b) the definition of metrics that serve as proxies for quantum kernels with desirable properties; and (c) an evolutionary-guided search for quantum feature maps maximizing those metrics. When using this framework with sensible parameter choices, we obtain results suggesting that some previous studies may have reported overfitted outcomes. This shows that justifying the need for quantum mechanical properties might be beyond the actual scope of conventional classical tasks, since there is no clear quantum feature that contributes to the gain shown by some of the QML techniques most commonly applied in the literature.
Quantum-enhanced machine learning is a rapidly evolving field that aims to leverage the unique properties of quantum mechanics to enhance classical machine learning. However, the practical applicability of these methods remains an open question, particularly beyond the context of specifically crafted toy problems, and given the current limitations of quantum hardware. This study focuses on quantum kernel methods in the context of classification tasks. In particular, it examines the performance of quantum kernel estimation and quantum kernel training (QKT) in connection with two quantum feature mappings, namely, ZZFeatureMap and CovariantFeatureMap. Remarkably, these feature maps have been proposed in the literature under the conjecture of possible near-term quantum advantage and have shown promising performance in ad hoc datasets. This study aims to evaluate their versatility and generalization capabilities in a more general benchmark, encompassing both artificial and established reference datasets. Classical machine learning methods, specifically support vector machines and logistic regression, are also incorporated as baseline comparisons. Experimental results indicate that quantum methods exhibit varying performance across different datasets. Despite outperforming classical methods in ad hoc datasets, mixed results are obtained for the general case among standard classical benchmarks. The experimental data call into question a general added value of applying QKT optimization, for which the additional computational cost does not necessarily translate into improved classification performance. Instead, it is suggested that a careful choice of the quantum feature map in connection with proper hyperparameterization may prove more effective.
Abstract Since the entry of kernel theory in the field of quantum machine learning, quantum kernel methods (QKMs) have gained increasing attention with regard to both probing promising applications and delivering intriguing research insights. Benchmarking these methods is crucial to gain robust insights and to understand their practical utility. In this work, we present a comprehensive large-scale study examining QKMs based on fidelity quantum kernels (FQKs) and projected quantum kernels (PQKs) across a manifold of design choices. Our investigation encompasses both classification and regression tasks for five dataset families and 64 datasets, systematically comparing the use of FQKs and PQKs quantum support vector machines and kernel ridge regression. This resulted in over 20,000 models that were trained and optimized using a state-of-the-art hyperparameter search to ensure robust and comprehensive insights. We delve into the importance of hyperparameters on model performance scores and support our findings through rigorous correlation analyses. Additionally, we provide an in-depth analysis addressing the design freedom of PQKs and explore the underlying principles responsible for learning. Our goal is not to identify the best-performing model for a specific task but to uncover the mechanisms that lead to effective QKMs and reveal universal patterns.
The discovery of quantum algorithms offering provable advantages over the best known classical alternatives, together with the parallel ongoing revolution brought about by classical artificial intelligence, motivates a search for applications of quantum information processing methods to machine learning. Among several proposals in this domain, quantum kernel methods have emerged as particularly promising candidates. However, while some rigorous speedups on certain highly specific problems have been formally proven, only empirical proof-of-principle results have been reported so far for real-world datasets. Moreover, no systematic procedure is known, in general, to fine tune and optimize the performances of kernel-based quantum classification algorithms. At the same time, certain limitations such as kernel concentration effects-hindering the trainability of quantum classifiers-have also been recently pointed out. In this work, we propose several general-purpose optimization methods and best practices designed to enhance the practical usefulness of fidelity-based quantum classification algorithms. Specifically, we first describe a data pre-processing strategy that, by preserving the relevant relationships between data points when processed through quantum feature maps, substantially alleviates the effect of kernel concentration on structured datasets. We also introduce a classical post-processing method that, based on standard fidelity measures estimated on a quantum processor, yields non-linear decision boundaries in the feature Hilbert space, thus achieving the quantum counterpart of the radial basis functions technique that is widely employed in classical kernel methods. Finally, we apply the so-called quantum metric learning protocol to engineer and adjust trainable quantum embeddings, demonstrating substantial performance improvements on several paradigmatic real-world classification tasks.
In this article, we investigate the applicability of quantum machine learning for classification tasks using two quantum classifiers from the Qiskit Python environment: the variational quantum circuit and the quantum kernel estimator (QKE). We provide a first evaluation on the performance of these classifiers when using a hyperparameter search on six widely known and publicly available benchmark datasets and analyze how their performance varies with the number of samples on two artificially generated test classification datasets. As quantum machine learning is based on unitary transformations, this paper explores data structures and application fields that could be particularly suitable for quantum advantages. Hereby, this paper introduces a novel dataset based on concepts from quantum mechanics using the exponential map of a Lie algebra. This dataset will be made publicly available and contributes a novel contribution to the empirical evaluation of quantum supremacy. We further compared the performance of VQC and QKE on six widely applicable datasets to contextualize our results. Our results demonstrate that the VQC and QKE perform better than basic machine learning algorithms, such as advanced linear regression models (Ridge and Lasso). They do not match the accuracy and runtime performance of sophisticated modern boosting classifiers such as XGBoost, LightGBM, or CatBoost. Therefore, we conclude that while quantum machine learning algorithms have the potential to surpass classical machine learning methods in the future, especially when physical quantum infrastructure becomes widely available, they currently lag behind classical approaches. Our investigations also show that classical machine learning approaches have superior performance classifying datasets based on group structures, compared to quantum approaches that particularly use unitary processes. Furthermore, our findings highlight the significant impact of different quantum simulators, feature maps, and quantum circuits on the performance of the employed quantum estimators. This observation emphasizes the need for researchers to provide detailed explanations of their hyperparameter choices for quantum machine learning algorithms, as this aspect is currently overlooked in many studies within the field. To facilitate further research in this area and ensure the transparency of our study, we have made the complete code available in a linked GitHub repository.
Quantum kernel methods are the most concrete near-term proposal for quantum machine learning on classical data, yet published benchmarks disagree on whether they outperform tuned classical kernels, and the disagreement cannot be resolved from the literature because the comparisons are not protocol-controlled. We argue that the obstacle is one of measurement, not theory: the usual setup fixes the quantum feature map's data scaling while grid-searching the classical kernel's hyperparameters, recomputes nothing per fold, and reports cost in simulator time. We therefore treat the comparison itself as the object of study and impose three controls. The bandwidth is tuned on the same grid budget as the classical γ, the kernel is recomputed inside every fold from fold-fitted preprocessing, and cost is counted in hardware circuit evaluations. On top of these we run a factorial design of 324 cells in which dimensionality, training size, class separability and data structure are explicit covariates, together with four standard datasets, a shots-versus-gate-noise ablation, and a cost sweep. The dominant effect we measure is not quantum: tuning the feature map's bandwidth is worth 0.120 accuracy on synthetic data and 0.339 on real data, exceeding every quantum-classical difference we observe. Tuned quantum kernels are then statistically indistinguishable from a tuned RBF on all four real datasets. The advantage on synthetic data is confined almost entirely to data generated by the feature map itself, and it shrinks with dimensionality (-0.0045 per qubit) while growing with sample size. Finite sampling costs an order of magnitude more accuracy than gate and readout noise, and a fidelity kernel needs 5×107 circuits at n = 10,000. Protocol variance is thus sufficient to explain the field's contradictory conclusions.