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:2312.01655

Metric Learning with Quantum Feature Spaces

Sharma, Vinayak, Sen, Lord, Padhy, Ashish, Behera, Sourav, De Luca, Gennaro, Mukherjee, Shyamapada, Aviral Shrivastava

Published Dec 4, 2023• 1 citation
arXivHTMLPDF

Abstract

The $2^N$ dimensions of Hilbert space give quantum metric learning exponentially many mutually orthogonal states with which to separate classes, but the same exponential dimension also degrades its optimization landscape and the tractability of evaluating its metric. By letting a classical network predict all gate parameters of a fixed ansatz, we isolate the reachable state manifold and its fidelity metric. We prove that this separability is already present on the unentangled manifold $\mathbb{M}_{\text{unentangled}} = (\mathbb{CP}^1)^N$, which saturates the $2^N$ maximal orthogonal capacity needed for classification, while entanglement adds states only at intermediate separations, which are costly to resolve under finite sampling. The same size carries costs. Global fidelity causes exponential gradient collapse across both unentangled ($Θ(c^{-N})$) and entangled ($Θ(4^{-N})$) manifolds; as fidelities concentrate, the fidelity signal-to-noise ratio at a fixed shot budget also decays exponentially in $N$; and because $\mathbb{M}_{\text{unentangled}}$ already holds $2^N$ orthogonal states, expressive, unconstrained encoders under contrastive losses reach trivial minima by \emph{geometric shattering} even without entanglement. We introduce Quantum Projective Metric Learning (QPMeL), an exact $O(N)$ closed-form kernel on $\mathbb{M}_{\text{unentangled}}$ that circumvents circuit simulation. QPMeL resolves gradient collapse via a curriculum annealing from a $1$-local observable regime (restoring $Θ(1/N^2)$ polynomial variance) to the global product kernel. Empirically, QPMeL as a complete design exceeds eight and matches two of 11 published quantum few-shot designs.

DOI
10.48550/arxiv.2312.01655

Related papers

LieTrunc-QNN: Lie Algebra Truncation and Quantum Expressivity Phase Transition from LiePrune to Provably Stable Quantum Neural Networks

Haijian Shao, Dalong A. Zhao, Xing Deng +2 more

Quantum Machine Learning (QML) is fundamentally limited by two challenges: barren plateaus (exponentially vanishing gradients) and the fragility of parameterized quantum circuits under noise. Despite extensive empirical studies, a unified theoretical framework remains lacking. We introduce LieTrunc-QNN, an algebraic-geometric framework that characterizes trainability via Lie-generated dynamics. Parameterized quantum circuits are modeled as Lie subalgebras of u(2^n), whose action induces a Riemannian manifold of reachable quantum states. Expressivity is reinterpreted as intrinsic manifold dimension and geometry. We establish a geometric capacity-plateau principle: increasing effective dimension leads to exponential gradient suppression due to concentration of measure. By restricting to structured Lie subalgebras (LieTrunc), the manifold is contracted, preventing concentration and preserving non-degenerate gradients. We prove two main results: (1) a trainability lower bound for LieTrunc-QNN, and (2) that the Fubini-Study metric rank is bounded by the algebraic span of generators, showing expressivity is governed by structure rather than parameter count. Compact Lie subalgebras also provide inherent robustness to perturbations. Importantly, we establish a polynomial trainability regime where gradient variance decays polynomially instead of exponentially. Experiments (n=2-6) validate the theory: LieTrunc-QNN maintains stable gradients and high effective dimension, while random truncation leads to metric rank collapse. At n=6, full metric rank is preserved (rank=16). Results support a scaling law between gradient variance and effective dimension. This work provides a unified geometric framework for QNN design, linking Lie algebra, manifold geometry, and optimization.

arXiv:2604.02697•Apr 3, 2026•0 citations

Geometric Latent Space Tomography with Metric-Preserving Autoencoders

S. M. Yousuf Iqbal Tomal, Abdullah Al Shafin

Quantum state tomography faces exponential scaling with system size, while recent neural network approaches achieve polynomial scaling at the cost of losing the geometric structure of quantum state space. We introduce geometric latent space tomography, combining classical neural encoders with parameterized quantum circuit decoders trained via a metric-preservation loss that enforces proportionality between latent Euclidean distances and quantum Bures geodesics. On two-qubit mixed states with purity 0.85--0.95 representing NISQ-era decoherence, we achieve high-fidelity reconstruction (mean fidelity $F = 0.942 \pm 0.03$) with an interpretable 20-dimensional latent structure. Critically, latent geodesics exhibit strong linear correlation with Bures distances (Pearson $r = 0.88$, $R^2 = 0.78$), preserving 78\% of quantum metric structure. Geometric analysis reveals intrinsic manifold dimension 6.35 versus 20 ambient dimensions and measurable local curvature ($κ= 0.011 \pm 0.006$), confirming non-trivial Riemannian geometry with $O(d^2)$ computational advantage over $O(4^n)$ density matrix operations. Unlike prior neural tomography, our geometry-aware latent space enables direct state discrimination, fidelity estimation from Euclidean distances, and interpretable error manifolds for quantum error mitigation without repeated full tomography, providing critical capabilities for NISQ devices with limited coherence times.

quant-pharXiv:2512.15801•Dec 18, 2025•0 citations

A Fundamental Duality in Quantum Machine Learning: Trainability vs. Generalization via Dynamical Lie Algebras

Ayad Alhusseiny

Quantum learnability is an algebraic invariant. While optimization and statistical generalization have long been analyzed through disconnected mathematical formalisms in quantum machine learning, both phenomena emerge as dual geometric projections of the dynamical Lie algebra (DLA) dimension g = Lie{-i H_k} ⊆ su(2^n); across any parameterized circuit family, polynomial trainability (Var[∇C] ≥ Ω(1/poly(n))) and polynomial generalizability (M*(ε) ≤ O(poly(n)/ε^2)) are strictly equivalent and hold if and only if dim(g_n) ∈ O(poly(n)). For any observable with adjoint orbit O_O, the operational product satisfies Var[∇C] · M*(ε) = Θ(dim(O_O) / (dim(g_n) ε^2)), which collapses to the universal, dimension-free invariant Θ(1/ε^2) for all generic, non-degenerate observables, identically independent of system size n and algebraic dimension dim(g_n). Unconstrained expressivity (dim(g_n) ~ 4^n) triggers a simultaneous catastrophe: exponential gradient vanishing accompanied by exponential sample complexity explosion. Subalgebra confinement cures both failures. By restricting generators to permutation-equivariant or symmetry-protected subalgebras, the circuit enforces dim(g_n) ≤ O(poly(n)) and secures non-vanishing gradient flow alongside statistical sample efficiency. Exact statevector simulations in PennyLane across n ∈ {2, 3, 4, 5, 6} corroborate this duality: hardware-efficient ansätze exhibit a 17.22× variance collapse, whereas symmetry-protected models sustain Θ(1) gradients and achieve twofold lower test error.

•Oct 23, 2026•0 citations

Empirical Evaluation of Quantum Kernel Methods for Classification: A Comparative Study with Classical Kernels

Ronak Bhalgami

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 $5\times10^{7}$ circuits at n = 10,000. Protocol variance is thus sufficient to explain the field's contradictory conclusions.

quant-pharXiv:2610.42029•Nov 2, 2026•0 citations

A Unified Classification Framework with Quantum Metric Learning

Nhat A. Nghiem, Samuel Yen-Chi Chen, Tzu-Chieh Wei

Recently, Lloyd et al. [arXiv:2001.03622] introduced \textit{quantum metric learning} as a powerful framework for quantum supervised learning. Such strategy focuses on the training of the quantum embedding, with a goal of `well-separation' of the embedded data in the Hilbert space $\mathcal{H}$. Motivated by such an idea, we present a metric-based framework for supervised learning with trainable quantum circuits. We introduce both explicit and implicit approaches, based on the \textit{metric} defined on the Hilbert space $\mathcal{H}$, to classify data. In particular, with the implicit approach, the number of separated classes (or their labels) in supervised learning problems can be arbitrarily high (with respect to the number of given qubits, or more precisely, the number of measurement), which is an improvement over current quantum supervised learning models, and this approach requires relatively small number of training points. With the explicit approach, we establish an equivalence between our metric-based framework and other quantum supervised learning models. Such equivalence provides a unification of supervised learning frameworks on quantum computer. We also demonstrate our framework by performing both noise-free and noisy numerical simulations and by running the classification on real IBM Q devices.

•Oct 25, 2020•6 citations

Quantum machine learning models for graphs

Frédéric Sauvage, Pranav Kalidindi, Frederic Rapp +1 more

Geometric Machine Learning (GML) successes have been achieved through the thorough study and design of new equivariant neural networks. In comparison, geometric quantum machine learning (GQML) models lack such a detailed understanding and, despite already several proposals, a unifying perspective on their design remains elusive. In this work, we focus on GQML models for graph problems that showcase a lot of structure and still remain frontier in machine learning. For the case when n-node graphs are encoded in n-qubit states, we provide a comprehensive characterization of their constituents. Taken together, these furnish us with a toolbox for the design of quantum graph models, and we further probe its benefits including the natural integration with classical models, generalization of known GQML models (sometimes extending their expressivity at virtually no cost), and straightforward classical pre-training strategies. The latter two features are demonstrated in dedicated numerical experiments.

quant-pharXiv:2607.00698•Jul 2, 2026•1 citation
The 2N2^N2N dimensions of Hilbert space give quantum metric learning exponentially many mutually orthogonal states with which to separate classes, but the same exponential dimension also degrades its optimization landscape and the tractability of evaluating its metric. By letting a classical network predict all gate parameters of a fixed ansatz, we isolate the reachable state manifold and its fidelity metric. We prove that this separability is already present on the unentangled manifold Munentangled=(CP1)N\mathbb{M}_{\text{unentangled}} = (\mathbb{CP}^1)^NMunentangled​=(CP1)N, which saturates the 2N2^N2N maximal orthogonal capacity needed for classification, while entanglement adds states only at intermediate separations, which are costly to resolve under finite sampling. The same size carries costs. Global fidelity causes exponential gradient collapse across both unentangled (Θ(c−N)Θ(c^{-N})Θ(c−N)) and entangled (Θ(4−N)Θ(4^{-N})Θ(4−N)) manifolds; as fidelities concentrate, the fidelity signal-to-noise ratio at a fixed shot budget also decays exponentially in NNN; and because Munentangled\mathbb{M}_{\text{unentangled}}Munentangled​ already holds 2N2^N2N orthogonal states, expressive, unconstrained encoders under contrastive losses reach trivial minima by \emph{geometric shattering} even without entanglement. We introduce Quantum Projective Metric Learning (QPMeL), an exact O(N)O(N)O(N) closed-form kernel on Munentangled\mathbb{M}_{\text{unentangled}}Munentangled​ that circumvents circuit simulation. QPMeL resolves gradient collapse via a curriculum annealing from a 111-local observable regime (restoring Θ(1/N2)Θ(1/N^2)Θ(1/N2) polynomial variance) to the global product kernel. Empirically, QPMeL as a complete design exceeds eight and matches two of 11 published quantum few-shot designs.
Quantum Machine Learning (QML) is fundamentally limited by two challenges: barren plateaus (exponentially vanishing gradients) and the fragility of parameterized quantum circuits under noise. Despite extensive empirical studies, a unified theoretical framework remains lacking. We introduce LieTrunc-QNN, an algebraic-geometric framework that characterizes trainability via Lie-generated dynamics. Parameterized quantum circuits are modeled as Lie subalgebras of u(2^n), whose action induces a Riemannian manifold of reachable quantum states. Expressivity is reinterpreted as intrinsic manifold dimension and geometry. We establish a geometric capacity-plateau principle: increasing effective dimension leads to exponential gradient suppression due to concentration of measure. By restricting to structured Lie subalgebras (LieTrunc), the manifold is contracted, preventing concentration and preserving non-degenerate gradients. We prove two main results: (1) a trainability lower bound for LieTrunc-QNN, and (2) that the Fubini-Study metric rank is bounded by the algebraic span of generators, showing expressivity is governed by structure rather than parameter count. Compact Lie subalgebras also provide inherent robustness to perturbations. Importantly, we establish a polynomial trainability regime where gradient variance decays polynomially instead of exponentially. Experiments (n=2-6) validate the theory: LieTrunc-QNN maintains stable gradients and high effective dimension, while random truncation leads to metric rank collapse. At n=6, full metric rank is preserved (rank=16). Results support a scaling law between gradient variance and effective dimension. This work provides a unified geometric framework for QNN design, linking Lie algebra, manifold geometry, and optimization.
Quantum state tomography faces exponential scaling with system size, while recent neural network approaches achieve polynomial scaling at the cost of losing the geometric structure of quantum state space. We introduce geometric latent space tomography, combining classical neural encoders with parameterized quantum circuit decoders trained via a metric-preservation loss that enforces proportionality between latent Euclidean distances and quantum Bures geodesics. On two-qubit mixed states with purity 0.85--0.95 representing NISQ-era decoherence, we achieve high-fidelity reconstruction (mean fidelity F=0.942±0.03F = 0.942 \pm 0.03F=0.942±0.03) with an interpretable 20-dimensional latent structure. Critically, latent geodesics exhibit strong linear correlation with Bures distances (Pearson r=0.88r = 0.88r=0.88, R2=0.78R^2 = 0.78R2=0.78), preserving 78\% of quantum metric structure. Geometric analysis reveals intrinsic manifold dimension 6.35 versus 20 ambient dimensions and measurable local curvature (κ=0.011±0.006κ= 0.011 \pm 0.006κ=0.011±0.006), confirming non-trivial Riemannian geometry with O(d2)O(d^2)O(d2) computational advantage over O(4n)O(4^n)O(4n) density matrix operations. Unlike prior neural tomography, our geometry-aware latent space enables direct state discrimination, fidelity estimation from Euclidean distances, and interpretable error manifolds for quantum error mitigation without repeated full tomography, providing critical capabilities for NISQ devices with limited coherence times.
Quantum learnability is an algebraic invariant. While optimization and statistical generalization have long been analyzed through disconnected mathematical formalisms in quantum machine learning, both phenomena emerge as dual geometric projections of the dynamical Lie algebra (DLA) dimension g = Lie{-i H_k} ⊆ su(2^n); across any parameterized circuit family, polynomial trainability (Var[∇C] ≥ Ω(1/poly(n))) and polynomial generalizability (M*(ε) ≤ O(poly(n)/ε^2)) are strictly equivalent and hold if and only if dim(g_n) ∈ O(poly(n)). For any observable with adjoint orbit O_O, the operational product satisfies Var[∇C] · M*(ε) = Θ(dim(O_O) / (dim(g_n) ε^2)), which collapses to the universal, dimension-free invariant Θ(1/ε^2) for all generic, non-degenerate observables, identically independent of system size n and algebraic dimension dim(g_n). Unconstrained expressivity (dim(g_n) ~ 4^n) triggers a simultaneous catastrophe: exponential gradient vanishing accompanied by exponential sample complexity explosion. Subalgebra confinement cures both failures. By restricting generators to permutation-equivariant or symmetry-protected subalgebras, the circuit enforces dim(g_n) ≤ O(poly(n)) and secures non-vanishing gradient flow alongside statistical sample efficiency. Exact statevector simulations in PennyLane across n ∈ {2, 3, 4, 5, 6} corroborate this duality: hardware-efficient ansätze exhibit a 17.22× variance collapse, whereas symmetry-protected models sustain Θ(1) gradients and achieve twofold lower test error.
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 γγγ, 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 5×1075\times10^{7}5×107 circuits at n = 10,000. Protocol variance is thus sufficient to explain the field's contradictory conclusions.
Recently, Lloyd et al. [arXiv:2001.03622] introduced \textit{quantum metric learning} as a powerful framework for quantum supervised learning. Such strategy focuses on the training of the quantum embedding, with a goal of `well-separation' of the embedded data in the Hilbert space H\mathcal{H}H. Motivated by such an idea, we present a metric-based framework for supervised learning with trainable quantum circuits. We introduce both explicit and implicit approaches, based on the \textit{metric} defined on the Hilbert space H\mathcal{H}H, to classify data. In particular, with the implicit approach, the number of separated classes (or their labels) in supervised learning problems can be arbitrarily high (with respect to the number of given qubits, or more precisely, the number of measurement), which is an improvement over current quantum supervised learning models, and this approach requires relatively small number of training points. With the explicit approach, we establish an equivalence between our metric-based framework and other quantum supervised learning models. Such equivalence provides a unification of supervised learning frameworks on quantum computer. We also demonstrate our framework by performing both noise-free and noisy numerical simulations and by running the classification on real IBM Q devices.
Geometric Machine Learning (GML) successes have been achieved through the thorough study and design of new equivariant neural networks. In comparison, geometric quantum machine learning (GQML) models lack such a detailed understanding and, despite already several proposals, a unifying perspective on their design remains elusive. In this work, we focus on GQML models for graph problems that showcase a lot of structure and still remain frontier in machine learning. For the case when n-node graphs are encoded in n-qubit states, we provide a comprehensive characterization of their constituents. Taken together, these furnish us with a toolbox for the design of quantum graph models, and we further probe its benefits including the natural integration with classical models, generalization of known GQML models (sometimes extending their expressivity at virtually no cost), and straightforward classical pre-training strategies. The latter two features are demonstrated in dedicated numerical experiments.