pith. sign in

arxiv: 2606.13244 · v1 · pith:PD63CVSYnew · submitted 2026-06-11 · 🪐 quant-ph

Coupling-Grouped XY-QAOA for Joint Anomaly-Feature Selection

Pauli Taipale This is my paper

Pith reviewed 2026-06-27 06:26 UTC · model grok-4.3

classification 🪐 quant-ph
keywords joint anomaly-feature selectionXY-QAOAconstraint-preserving QAOAbipartite selectionquantum hardware benchmarkscalibration error sensitivitygrouped anglesanomaly detection
0
0 comments X

The pith

Joint sample-feature selection keeps calibration-error sensitivity constant and is solved by Coupling-Grouped XY-QAOA that cuts circuit depth 45.9-61.3%.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper formulates anomaly and feature selection under fixed budgets as one coupled optimization problem instead of selecting features first then samples. In its formal model the joint version makes calibration-error sensitivity independent of sample count while feature-first ordering makes sensitivity grow linearly with samples. Coupling-Grouped XY-QAOA is introduced as a grouped-angle, constraint-preserving solver for the joint problem. On IBM Heron hardware the method reduces depth by 45.9-61.3 percent and two-qubit gates by 2.6-5.2 percent relative to Qiskit level-3 optimization, enabling 64-qubit p=2 and 36-qubit p=3 executions that retain feasible sectors. Fixed-angle and warm-start variants also produce lower-energy feasible samples than random or classical baselines in the tested regimes.

Core claim

Selecting anomalous samples and explanatory features under fixed budgets defines a coupled constrained-optimization problem. Sequential feature-first selection ranks features before choosing samples, which can overlook features whose utility depends on which samples are selected, especially when scores are calibrated from reference data that may be limited, noisy, or drifting. We instead formulate the task as joint sample-feature selection under the same fixed counts. In the analyzed formal model, calibration-error sensitivity grows linearly with the number of samples for feature-first ordering but stays constant for joint selection. We introduce Coupling-Grouped XY-QAOA, a constraint-preser

What carries the argument

Coupling-Grouped XY-QAOA, a constraint-preserving grouped-angle variant of XY-QAOA that encodes bipartite sample-feature selection while reducing mixer depth.

If this is right

  • Enables the largest reported width-depth configurations for constraint-preserving bipartite-selection QAOA: 64 qubits at p=2 and 36 qubits at p=3 with feasible-sector retention.
  • Fixed-angle runs yield lower-energy feasible samples than matched random-feasible sampling across 36-64 qubits.
  • Warm starts reduce the gap to strict-feasible classical references by 57.5-80.5 percent.
  • Near-budget repair matches the sparse classical reference at 36 qubits.
  • Problem-structured angle grouping improves performance over same-depth XY-QAOA and matched-parameter randomization in noiseless simulations.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The constant sensitivity result suggests joint selection may remain stable on drifting reference data without repeated recalibration.
  • Grouped-angle grouping may transfer to other constrained QAOA mixers that encode multiple selection variables.
  • The reported 20-qubit p=5 runs retaining 63 percent valid samples indicate that feasible-sector retention scales with problem structure rather than depth alone.

Load-bearing premise

The formal model correctly shows calibration-error sensitivity grows linearly with sample count for feature-first ordering but stays constant for joint selection.

What would settle it

Measure selection accuracy under controlled calibration noise while increasing the number of samples; accuracy should degrade linearly in feature-first runs but remain flat in joint runs.

Figures

Figures reproduced from arXiv: 2606.13244 by Pauli Taipale.

Figure 1
Figure 1. Figure 1: Logical CG-XY-QAOA circuit for exact-budget bipartite selection ( [PITH_FULL_IMAGE:figures/full_fig_p014_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Fully grouped cost-plus-mixer CG-XY-QAOA gives positive same-depth noiseless score [PITH_FULL_IMAGE:figures/full_fig_p020_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Sparse-cost scaling across hardware targets. Hardware-aligned cost layouts reduce submitted two-qubit resources across the IQM Emerald square-lattice and IBM Heron R3 heavy￾hex targets. Each point is the median over 20 transpiler seeds. Bands show interquartile ranges. Decision qubits N + D 40 60 80 100 Submitted depth Circuit depth p = 2 Decision qubits N + D 100 150 200 250 Submitted 2Q gates CZ-basis cl… view at source ↗
Figure 4
Figure 4. Figure 4: Implementation-path scaling on hardware-aligned sparse benchmark circuits. All curves use the same retained cost terms, preselected hardware patch, initial layout, basis initialization, bilinear-CG parameterization, representative fixed angles, physical backend, and transpiler seed. The curves compare Opt-3 on the standard CZ-basis target, Opt-3 on the fractional-gate target, and paths that add Block XY fu… view at source ↗
read the original abstract

Selecting anomalous samples and explanatory features under fixed budgets defines a coupled constrained-optimization problem. Sequential feature-first selection ranks features before choosing samples, which can overlook features whose utility depends on which samples are selected, especially when scores are calibrated from reference data that may be limited, noisy, or drifting. We instead formulate the task as joint sample-feature selection under the same fixed counts. In the analyzed formal model, calibration-error sensitivity grows linearly with the number of samples for feature-first ordering but stays constant for joint selection. We introduce Coupling-Grouped XY-QAOA, a constraint-preserving grouped-angle variant for the resulting optimization problem. On matched sparse IBM Heron R3 benchmarks, a hardware-aware implementation reduces circuit depth by 45.9%-61.3% and two-qubit gates by 2.6%-5.2% relative to Qiskit optimization level 3 on the CZ-basis target. It enables, to our knowledge, the largest reported width-depth configurations for constraint-preserving bipartite-selection QAOA hardware executions with feasible-sector retention: 64 qubits at p=2 and 36 qubits at p=3. The 20-qubit p=5 runs retain 63% valid samples. Across 36-64 qubits, fixed-angle runs yield lower-energy feasible samples than matched random-feasible sampling. Warm starts reduce the gap to strict-feasible classical references by 57.5%-80.5%, and near-budget repair matches the sparse classical reference at 36 qubits. Benchmarks show gains in balanced fixed-budget regimes, and noiseless simulations show that problem-structured angle grouping improves over same-depth XY-QAOA and matched-parameter, type-preserving randomization controls. Overall, the results support calibrated joint selection and hardware-realizable constrained-mixer execution in the tested regimes.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 3 minor

Summary. The paper claims that joint sample-feature selection under fixed budgets outperforms sequential feature-first ordering in calibration-error sensitivity (linear growth vs constant), introduces Coupling-Grouped XY-QAOA as a constraint-preserving grouped-angle QAOA variant, and reports hardware benchmarks on IBM Heron R3 showing 45.9%-61.3% depth and 2.6%-5.2% two-qubit gate reductions relative to Qiskit level 3, enabling record configurations (64 qubits at p=2, 36 qubits at p=3) with feasible-sector retention and competitive energy/repair performance versus classical references.

Significance. If the formal model and benchmarks hold, the work provides a concrete demonstration of hardware-realizable constrained QAOA for bipartite selection in anomaly detection, with efficiency gains that expand feasible width-depth regimes and a theoretical distinction favoring joint over sequential selection under calibration noise. The explicit comparisons to random-feasible sampling, warm starts, and near-budget repair add practical value.

minor comments (3)
  1. Abstract: the reported depth/gate reductions and qubit records are presented without accompanying error bars, exact instance sizes, or sparsity parameters for the 'matched sparse' benchmarks, which would aid verification of the 45.9%-61.3% and 2.6%-5.2% figures.
  2. Abstract: the formal model result on calibration-error sensitivity (linear vs constant) is stated as an analyzed outcome but lacks a pointer to the specific equations or assumptions used in the derivation.
  3. The 20-qubit p=5 retention of 63% valid samples and the 57.5%-80.5% warm-start gap reduction are useful metrics but would benefit from explicit definition of 'valid samples' and the classical reference baseline in the abstract.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive summary, significance assessment, and recommendation of minor revision. The report correctly captures the core claims regarding joint versus sequential selection under calibration noise, the Coupling-Grouped XY-QAOA formulation, and the reported hardware depth/gate reductions plus feasible-sector results on IBM Heron. No specific major comments appear in the provided report, so we have no point-by-point items to address.

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper's derivation chain consists of an analyzed formal model whose sensitivity result is stated as an output of that model, followed by an empirical QAOA construction whose performance claims (depth/gate reductions, width-depth records, feasible retention) are obtained from direct IBM Heron hardware executions and noiseless simulations. No equation or parameter is fitted to a subset and then renamed as a prediction; no self-citation is invoked as a uniqueness theorem or load-bearing premise; the joint-selection advantage is not defined in terms of itself. The reported benchmarks therefore remain independent of the paper's own fitted values or prior-author citations.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract provides no explicit details on free parameters, axioms, or invented entities; full manuscript required for identification.

pith-pipeline@v0.9.1-grok · 5848 in / 1221 out tokens · 33755 ms · 2026-06-27T06:26:00.237833+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

66 extracted references · 52 canonical work pages

  1. [1]

    Hammer and Sergiu Rudeanu

    Peter L. Hammer and Sergiu Rudeanu. Pseudo-Boolean programming.Operations Research, 17(2):233–261, 1969. doi: 10.1287/opre.17.2.233

    work page doi:10.1287/opre.17.2.233 1969

  2. [2]

    (2014) Ising formulations of many NP problems

    Andrew Lucas. Ising formulations of many NP problems.Frontiers in Physics, 2:5, 2014. doi: 10.3389/fphy.2014.00005

    work page doi:10.3389/fphy.2014.00005 2014

  3. [3]

    XY mixers: Analytical and numerical results for the quantum alternating operator ansatz.Physical Review A, 101(1):012320, 2020

    Zhihui Wang, Nicholas C Rubin, Jason M Dominy, and Eleanor G Rieffel. XY mixers: Analytical and numerical results for the quantum alternating operator ansatz.Physical Review A, 101(1):012320, 2020. doi: 10.1103/PhysRevA.101.012320

    work page doi:10.1103/physreva.101.012320 2020

  4. [4]

    Lotshaw, James Ostrowski, Travis S

    Rebekah Herrman, Phillip C. Lotshaw, James Ostrowski, Travis S. Humble, and George Siopsis. Multi-angle quantum approximate optimization algorithm.Scientific Reports, 12 (1), 2022. doi: 10.1038/s41598-022-10555-8

    work page doi:10.1038/s41598-022-10555-8 2022

  5. [5]

    A quantum approximate optimization algorithm.arXiv preprint arXiv:1411.4028, 2014

    Edward Farhi, Jeffrey Goldstone, and Sam Gutmann. A quantum approximate optimization algorithm.arXiv preprint arXiv:1411.4028, 2014. URL https://arxiv.org/abs/1411.4 028

    Pith/arXiv arXiv 2014

  6. [6]

    From the quantum approximate optimization algorithm to a quantum alternating operator ansatz.Algorithms, 12(2):34, 2019

    Stuart Hadfield, Zhihui Wang, Bryan O’Gorman, Eleanor G Rieffel, Davide Venturelli, and Rupak Biswas. From the quantum approximate optimization algorithm to a quantum alternating operator ansatz.Algorithms, 12(2):34, 2019. doi: 10.3390/a12020034

    work page doi:10.3390/a12020034 2019

  7. [7]

    Coherence in spontaneous radiation processes.Physical Review, 93(1): 99–110, 1954

    Robert H Dicke. Coherence in spontaneous radiation processes.Physical Review, 93(1): 99–110, 1954. doi: 10.1103/PhysRev.93.99

    work page doi:10.1103/physrev.93.99 1954

  8. [8]

    Deterministic preparation of Dicke states

    Andreas B¨ artschi and Stephan Eidenbenz. Deterministic preparation of Dicke states. In Fundamentals of Computation Theory: 22nd International Symposium, FCT 2019, volume 11651, pages 126–139. Springer, 2019. doi: 10.1007/978-3-030-25027-0 9

    work page doi:10.1007/978-3-030-25027-0 2019

  9. [9]

    Grover mixers for QAOA: Shifting complexity from mixer design to state preparation

    Andreas B¨ artschi and Stephan Eidenbenz. Grover mixers for QAOA: Shifting complexity from mixer design to state preparation. In2020 IEEE International Conference on Quantum Computing and Engineering (QCE), pages 72–82. IEEE, 2020. doi: 10.1109/QCE49297.202 0.00020

    work page doi:10.1109/qce49297.202 2020

  10. [10]

    Egger, Jakub Mareˇ cek, and Stefan Woerner

    Daniel J. Egger, Jakub Mareˇ cek, and Stefan Woerner. Warm-starting quantum optimization. Quantum, 5:479, 2021. doi: 10.22331/q-2021-06-17-479

    work page doi:10.22331/q-2021-06-17-479 2021

  11. [11]

    Harrigan et al

    Matthew P. Harrigan et al. Quantum approximate optimization of non-planar graph problems on a planar superconducting processor.Nature Physics, 17(3):332–336, 2021. doi: 10.1038/s41567-020-01105-y

    work page doi:10.1038/s41567-020-01105-y 2021

  12. [12]

    Ebadi et al

    S. Ebadi et al. Quantum optimization of maximum independent set using Rydberg atom arrays.Science, 376(6598):1209–1215, 2022. doi: 10.1126/science.abo6587

    work page doi:10.1126/science.abo6587 2022

  13. [13]

    Lukin, Sheng-Tao Wang, and Hannes Pichler

    Minh-Thi Nguyen, Jin-Guo Liu, Jonathan Wurtz, Mikhail D. Lukin, Sheng-Tao Wang, and Hannes Pichler. Quantum optimization with arbitrary connectivity using Rydberg atom arrays.PRX Quantum, 4(1):010316, 2023. doi: 10.1103/PRXQuantum.4.010316

    work page doi:10.1103/prxquantum.4.010316 2023

  14. [14]

    Alignment between initial state and mixer improves QAOA performance for constrained optimization.npj Quantum Information, 9(1):121, 2023

    Zichang He, Ruslan Shaydulin, Shouvanik Chakrabarti, Dylan Herman, Changhao Li, Yue Sun, and Marco Pistoia. Alignment between initial state and mixer improves QAOA performance for constrained optimization.npj Quantum Information, 9(1):121, 2023. doi: 10.1038/s41534-023-00787-5. 29

    work page doi:10.1038/s41534-023-00787-5 2023

  15. [15]

    Constrained quantum optimization for extractive summa- rization on a trapped-ion quantum computer.Scientific Reports, 12(1):17171, 2022

    Pradeep Niroula, Ruslan Shaydulin, Romina Yalovetzky, Pierre Minssen, Dylan Herman, Shaohan Hu, and Marco Pistoia. Constrained quantum optimization for extractive summa- rization on a trapped-ion quantum computer.Scientific Reports, 12(1):17171, 2022. doi: 10.1038/s41598-022-20853-w

    work page doi:10.1038/s41598-022-20853-w 2022

  16. [16]

    Quantum approximate multi-objective optimization.Nature Computational Science, 5(12):1168–1177, 2025

    Ayse Kotil et al. Quantum approximate multi-objective optimization.Nature Computational Science, 5(12):1168–1177, 2025. doi: 10.1038/s43588-025-00873-y

    work page doi:10.1038/s43588-025-00873-y 2025

  17. [17]

    Short-depth QAOA circuits and quantum annealing on higher-order Ising models.npj Quantum Information, 10(30),

    Elijah Pelofske, Andreas B¨ artschi, and Stephan Eidenbenz. Short-depth QAOA circuits and quantum annealing on higher-order Ising models.npj Quantum Information, 10(30),

  18. [18]

    127-qubit resource-matched comparison of QAOA against classical heuristics

    doi: 10.1038/s41534-024-00825-w. 127-qubit resource-matched comparison of QAOA against classical heuristics

    work page doi:10.1038/s41534-024-00825-w

  19. [19]

    QAOA with n·p≥ 200

    Ruslan Shaydulin and Marco Pistoia. QAOA with n·p≥ 200. In2023 IEEE International Conference on Quantum Computing and Engineering (QCE), pages 1074–1077, 2023. doi: 10.1109/QCE57702.2023.00121

    work page doi:10.1109/qce57702.2023.00121 2023

  20. [20]

    Quantum approximate optimization algorithm with sparsified phase operator

    Xiaoyuan Liu, Ruslan Shaydulin, and Ilya Safro. Quantum approximate optimization algorithm with sparsified phase operator. In2022 IEEE International Conference on Quantum Computing and Engineering (QCE), pages 133–141. IEEE, 2022. doi: 10.1109/QC E53715.2022.00032

    work page doi:10.1109/qc 2022

  21. [21]

    Hodson, Brendan Ruck, Hugh Ong, David Garvin, and Stefan Dulman

    Mark J. Hodson, Brendan Ruck, Hugh Ong, David Garvin, and Stefan Dulman. Portfolio rebalancing experiments using the quantum alternating operator ansatz.arXiv preprint arXiv:1911.05296, 2019. URLhttps://arxiv.org/abs/1911.05296

    arXiv 1911

  22. [22]

    Toward quantum gate-model heuristics for real-world planning problems.IEEE Transactions on Quantum Engineering, 1:1–16, 2020

    Tobias Stollenwerk, Stuart Hadfield, and Zhihui Wang. Toward quantum gate-model heuristics for real-world planning problems.IEEE Transactions on Quantum Engineering, 1:1–16, 2020. doi: 10.1109/TQE.2020.3030609

    work page doi:10.1109/tqe.2020.3030609 2020

  23. [23]

    Applying the quantum approximate optimization algorithm to the tail-assignment problem.Physical Review Applied, 14(3):034009, 2020

    Pontus Vikst˚ al, Mattias Gr¨ onkvist, Marika Svensson, Martin Andersson, G¨ oran Johansson, and Giulia Ferrini. Applying the quantum approximate optimization algorithm to the tail-assignment problem.Physical Review Applied, 14(3):034009, 2020. doi: 10.1103/Phys RevApplied.14.034009

    work page doi:10.1103/phys 2020

  24. [24]

    Isolation forest

    Fei Tony Liu, Kai Ming Ting, and Zhi-Hua Zhou. Isolation forest. In2008 Eighth IEEE International Conference on Data Mining, pages 413–422. IEEE, 2008. doi: 10.1109/ICDM .2008.17

    work page doi:10.1109/icdm 2008

  25. [25]

    LOF: identifying density-based local outliers

    Markus M Breunig, Hans-Peter Kriegel, Raymond T Ng, and J¨ org Sander. LOF: identifying density-based local outliers. InProceedings of the 2000 ACM SIGMOD international conference on Management of data, pages 93–104, 2000. doi: 10.1145/342009.335388

    work page doi:10.1145/342009.335388 2000

  26. [26]

    Anomaly detection: A survey.ACM Computing Surveys, 41(3):1–58, 2009

    Varun Chandola, Arindam Banerjee, and Vipin Kumar. Anomaly detection: A survey.ACM Computing Surveys, 41(3):1–58, 2009. doi: 10.1145/1541880.1541882

    work page doi:10.1145/1541880.1541882 2009

  27. [27]

    Explainable outlier detection: What, for whom and why?Machine Learning with Applications, 6:100172, 2021

    Jonas Herskind Sejr and Anna Schneider-Kamp. Explainable outlier detection: What, for whom and why?Machine Learning with Applications, 6:100172, 2021. doi: 10.1016/j.mlwa .2021.100172

    work page doi:10.1016/j.mlwa 2021

  28. [28]

    A survey on explainable anomaly detection.ACM Transactions on Knowledge Discovery from Data, 18(1):23:1–23:54, 2024

    Zhong Li, Yuxuan Zhu, and Matthijs van Leeuwen. A survey on explainable anomaly detection.ACM Transactions on Knowledge Discovery from Data, 18(1):23:1–23:54, 2024. doi: 10.1145/3609333

    work page doi:10.1145/3609333 2024

  29. [29]

    Contextual outlier interpretation

    Ninghao Liu, Donghwa Shin, and Xia Hu. Contextual outlier interpretation. InProceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence, pages 2461– 2467, 2018. doi: 10.24963/ijcai.2018/341. 30

    work page doi:10.24963/ijcai.2018/341 2018

  30. [30]

    A Stochastic Approximat ion Method

    Murray Rosenblatt. Remarks on a multivariate transformation.The Annals of Mathematical Statistics, 23(3):470–472, 1952. doi: 10.1214/aoms/1177729394

    work page doi:10.1214/aoms/1177729394 1952

  31. [31]

    The tight constant in the Dvoretzky-Kiefer-Wolfowitz inequality.The Annals of Probability, 18(3):1269–1283, 1990

    Pascal Massart. The tight constant in the Dvoretzky-Kiefer-Wolfowitz inequality.The Annals of Probability, 18(3):1269–1283, 1990. doi: 10.1214/aop/1176990746

    work page doi:10.1214/aop/1176990746 1990

  32. [32]

    Springer, 2022

    Vladimir Vovk, Alexander Gammerman, and Glenn Shafer.Algorithmic Learning in a Random World. Springer, 2022. doi: 10.1007/978-3-031-06649-8

    work page doi:10.1007/978-3-031-06649-8 2022

  33. [33]

    Tibshirani, and Larry Wasserman

    Jing Lei, Max G’Sell, Alessandro Rinaldo, Ryan J. Tibshirani, and Larry Wasserman. Distribution-free predictive inference for regression.Journal of the American Statistical Association, 113(523):1094–1111, 2018. doi: 10.1080/01621459.2017.1307116

    work page doi:10.1080/01621459.2017.1307116 2018

  34. [34]

    Cand` es, and Arnaud Durand

    Ery Arias-Castro, Emmanuel J. Cand` es, and Arnaud Durand. Detection of an anomalous cluster in a network.The Annals of Statistics, 39(1):278–304, 2011. doi: 10.1214/10-AOS839

    work page doi:10.1214/10-aos839 2011

  35. [35]

    Submatrix localization via message passing

    Bruce Hajek, Yihong Wu, and Jiaming Xu. Submatrix localization via message passing. Journal of Machine Learning Research, 18(186):1–52, 2018. URL https://jmlr.org/pap ers/v18/17-297.html

    2018

  36. [36]

    The maximum edge biclique problem is NP-complete.Discrete Applied Mathematics, 131(3):651–654, 2003

    Ren´ e Peeters. The maximum edge biclique problem is NP-complete.Discrete Applied Mathematics, 131(3):651–654, 2003. doi: 10.1016/S0166-218X(03)00333-0

    work page doi:10.1016/s0166-218x(03)00333-0 2003

  37. [37]

    West.Introduction to Graph Theory

    Douglas B. West.Introduction to Graph Theory. Prentice Hall, 2 edition, 2001. ISBN 9780130144003

    2001

  38. [38]

    Qiskit: An open-source framework for quantum computing

    Qiskit contributors. Qiskit: An open-source framework for quantum computing. Zenodo, 2019

    2019

  39. [39]

    Harun Bayraktar, Ali Charara, David Clark, Saul Cohen, Timothy Costa, Yao-Lung L. Fang, Yang Gao, Jack Guan, John Gunnels, Azzam Haidar, Andreas Hehn, Markus Hohnerbach, Matthew Jones, Tom Lubowe, Dmitry Lyakh, Shinya Morino, Paul Springer, Sam Stanwyck, Igor Terentyev, Satya Varadhan, Jonathan Wong, and Takuma Yamaguchi. cuQuantum SDK: A high-performance...

    work page doi:10.1109/qce57702.2023.00119 2023

  40. [40]

    Metropolis, A

    Nicholas Metropolis, Arianna W. Rosenbluth, Marshall N. Rosenbluth, Augusta H. Teller, and Edward Teller. Equation of state calculations by fast computing machines.The Journal of Chemical Physics, 21(6):1087–1092, 1953. doi: 10.1063/1.1699114

    work page doi:10.1063/1.1699114 1953

  41. [41]

    and Vecchi, M.P

    Scott Kirkpatrick, C. Daniel Gelatt, and Mario P. Vecchi. Optimization by simulated annealing.Science, 220(4598):671–680, 1983. doi: 10.1126/science.220.4598.671

    work page doi:10.1126/science.220.4598.671 1983

  42. [42]

    Kinetics of ising models

    Kyozi Kawasaki. Kinetics of ising models. In Cyril Domb and Melville S. Green, editors, Phase Transitions and Critical Phenomena, volume 2, pages 443–501. Academic Press, 1972

    1972

  43. [43]

    Interaction of Markov processes.Advances in Mathematics, 5(2):246–290,

    Frank Spitzer. Interaction of Markov processes.Advances in Mathematics, 5(2):246–290,

  44. [44]

    doi: 10.1016/0001-8708(70)90034-4

    work page doi:10.1016/0001-8708(70)90034-4

  45. [45]

    Johnson, and Gianluca Bontempi

    Andrea Dal Pozzolo, Olivier Caelen, Reid A. Johnson, and Gianluca Bontempi. Calibrating probability with undersampling for unbalanced classification. In2015 IEEE Symposium Series on Computational Intelligence, pages 159–166. IEEE, 2015. doi: 10.1109/ssci.2015.33

    work page doi:10.1109/ssci.2015.33 2015

  46. [46]

    Credit card fraud detection

    Machine Learning Group - ULB. Credit card fraud detection. Kaggle dataset, 2018. URL https://www.kaggle.com/datasets/mlg-ulb/creditcardfraud. Accessed 2026-01-30. 31

    2018

  47. [47]

    Realistic synthetic financial transactions for anti-money laundering models

    Erik Altman, Jovan Blanuˇ sa, Luc von Niederh¨ ausern, B´ eni Egressy, Andreea Anghel, and Kubilay Atasu. Realistic synthetic financial transactions for anti-money laundering models. InAdvances in Neural Information Processing Systems 36 (NeurIPS 2023), Datasets and Benchmarks Track, pages 29851–29874, 2023. doi: 10.52202/075280-1300. URL https://proceedi...

    work page doi:10.52202/075280-1300 2023

  48. [48]

    IBM transactions for anti money laundering (AML)

    Erik Altman. IBM transactions for anti money laundering (AML). Kaggle dataset, 2025. URL https://www.kaggle.com/datasets/ealtman2019/ibm-transactions-for-a nti-money-laundering-aml . Public Kaggle release of the synthetic AML transaction benchmark; Accessed 2026-04-21

    2025

  49. [49]

    Barkoutsos, Giacomo Nannicini, Anton Robert, Ivano Tavernelli, and Stefan Woerner

    Panagiotis Kl. Barkoutsos, Giacomo Nannicini, Anton Robert, Ivano Tavernelli, and Stefan Woerner. Improving variational quantum optimization using CVaR.Quantum, 4:256, 2020. doi: 10.22331/q-2020-04-20-256

    work page doi:10.22331/q-2020-04-20-256 2020

  50. [50]

    Sack and Daniel J

    Stefan H. Sack and Daniel J. Egger. Large-scale quantum approximate optimization on non-planar graphs with machine learning noise mitigation.Physical Review Research, 6(1): 013223, 2024. doi: 10.1103/PhysRevResearch.6.013223

    work page doi:10.1103/physrevresearch.6.013223 2024

  51. [51]

    Mixer-phaser ans¨ atze for quantum optimization with hard constraints.Quantum Machine Intelligence, 4(1):17, 2022

    Ryan LaRose, Eleanor Rieffel, and Davide Venturelli. Mixer-phaser ans¨ atze for quantum optimization with hard constraints.Quantum Machine Intelligence, 4(1):17, 2022. doi: 10.1007/s42484-022-00069-x

    work page doi:10.1007/s42484-022-00069-x 2022

  52. [52]

    Challenges and opportunities in quantum optimization.Nature Reviews Physics, 6(12):718–735, 2024

    Amira Abbas et al. Challenges and opportunities in quantum optimization.Nature Reviews Physics, 6(12):718–735, 2024. doi: 10.1038/s42254-024-00770-9

    work page doi:10.1038/s42254-024-00770-9 2024

  53. [53]

    Lov K. Grover. A fast quantum mechanical algorithm for database search. InProceedings of the 28th Annual ACM Symposium on Theory of Computing, pages 212–219, 1996. doi: 10.1145/237814.237866

    work page doi:10.1145/237814.237866 1996

  54. [54]

    Quantum amplitude amplifi- cation and estimation

    Gilles Brassard, Peter Høyer, Michele Mosca, and Alain Tapp. Quantum amplitude amplifi- cation and estimation. InQuantum Computation and Information, pages 53–74. American Mathematical Society, 2002. doi: 10.1090/conm/305/05215

    work page doi:10.1090/conm/305/05215 2002

  55. [55]

    A quantum algorithm for finding the minimum.arXiv preprint quant-ph/9607014, 1996

    Christoph D¨ urr and Peter Høyer. A quantum algorithm for finding the minimum.arXiv preprint quant-ph/9607014, 1996. URLhttps://arxiv.org/abs/quant-ph/9607014

    Pith/arXiv arXiv 1996

  56. [56]

    Coles, Lukasz Cincio, Jarrod R

    Martin Larocca, Supanut Thanasilp, Samson Wang, Kunal Sharma, Jacob Biamonte, Patrick J. Coles, Lukasz Cincio, Jarrod R. McClean, Zo¨ e Holmes, and Marco Cerezo. Barren plateaus in variational quantum computing.Nature Reviews Physics, 7:174–189, 2025. doi: 10.1038/s42254-025-00813-9

    work page doi:10.1038/s42254-025-00813-9 2025

  57. [57]

    McClean, Sergio Boixo, Vadim N

    Jarrod R. McClean, Sergio Boixo, Vadim N. Smelyanskiy, Ryan Babbush, and Hartmut Neven. Barren plateaus in quantum neural network training landscapes.Nature Communications, 9 (1):4812, 2018. doi: 10.1038/s41467-018-07090-4

    work page doi:10.1038/s41467-018-07090-4 2018

  58. [58]

    Marco Cerezo, Akira Sone, Tyler Volkoff, Lukasz Cincio, and Patrick J. Coles. Cost function dependent barren plateaus in shallow parametrized quantum circuits.Nature Communications, 12(1):1791, 2021. doi: 10.1038/s41467-021-21728-w

    work page doi:10.1038/s41467-021-21728-w 2021

  59. [59]

    Stuart Hadfield, Tad Hogg, and Eleanor G. Rieffel. Analytical framework for quantum alternating operator ans¨ atze.Quantum Science and Technology, 8(1):015017, 2023. doi: 10.1088/2058-9565/aca3ce. 32

    work page doi:10.1088/2058-9565/aca3ce 2023

  60. [60]

    2009 , month =

    St´ ephane Boucheron, G´ abor Lugosi, and Pascal Massart.Concentration Inequalities: A Nonasymptotic Theory of Independence. Oxford University Press, 2013. ISBN 9780199535255. doi: 10.1093/acprof:oso/9780199535255.001.0001

    work page doi:10.1093/acprof:oso/9780199535255.001.0001 2013

  61. [61]

    Probability inequalities for sums of bounded random variables.Journal of the American Statistical Association, 58(301):13–30, 1963

    Wassily Hoeffding. Probability inequalities for sums of bounded random variables.Journal of the American Statistical Association, 58(301):13–30, 1963. doi: 10.1080/01621459.1963.10 500830

    work page doi:10.1080/01621459.1963.10 1963

  62. [62]

    Empirical risk minimization for heavy- tailed losses.The Annals of Statistics, 43(6):2507–2536, 2015

    Christian Brownlees, Emilien Joly, and G´ abor Lugosi. Empirical risk minimization for heavy- tailed losses.The Annals of Statistics, 43(6):2507–2536, 2015. doi: 10.1214/15-AOS1350

    work page doi:10.1214/15-aos1350 2015

  63. [63]

    A simple sequentially rejective multiple test procedure.Scandinavian Journal of Statistics, 6(2):65–70, 1979

    Sture Holm. A simple sequentially rejective multiple test procedure.Scandinavian Journal of Statistics, 6(2):65–70, 1979. URLhttps://www.jstor.org/stable/4615733

    arXiv 1979

  64. [64]

    Rønnow, Zhihui Wang, Joshua Job, Sergio Boixo, Sergei V

    Troels F. Rønnow, Zhihui Wang, Joshua Job, Sergio Boixo, Sergei V. Isakov, David Wecker, John M. Martinis, Daniel A. Lidar, and Matthias Troyer. Defining and detecting quantum speedup.Science, 345(6195):420–424, 2014. doi: 10.1126/science.1252319. 33 Supplementary Material This supplement provides proofs, robustness analyses, and extended hardware results...

    work page doi:10.1126/science.1252319 2014

  65. [65]

    If the calibrated field has additive row and feature bias cWij =W ij +a i +b j, then bCj =C j +A+N b j, A:= NX i=1 ai, and bB(S, F) =B(S, F) +m X i∈S ai +k X j∈F bj. Thus row-only bias cancels from the sequential feature ranking, while feature bias is amplified by the pool size N in the sequential margin and enters the joint objective only through the fea...

  66. [66]

    Hence centered shared drift matters to the sequential feature ranking only through its nonzero row-sum component U

    If the calibrated field has rank-one drift cWij =W ij +u ivj, then bCj =C j +U v j, U:= NX i=1 ui, and bB(S, F) =B(S, F) + X i∈S ui ! X j∈F vj   . Hence centered shared drift matters to the sequential feature ranking only through its nonzero row-sum component U. A mean-shifted row factor behaves like feature bias after aggregation. Proof. Both claims ...