Papers
Topics
Authors
Recent
Search
2000 character limit reached

Out-of-distribution generalization for learning quantum dynamics

Published 21 Apr 2022 in quant-ph, cs.LG, and stat.ML | (2204.10268v3)

Abstract: Generalization bounds are a critical tool to assess the training data requirements of Quantum Machine Learning (QML). Recent work has established guarantees for in-distribution generalization of quantum neural networks (QNNs), where training and testing data are drawn from the same data distribution. However, there are currently no results on out-of-distribution generalization in QML, where we require a trained model to perform well even on data drawn from a different distribution to the training distribution. Here, we prove out-of-distribution generalization for the task of learning an unknown unitary. In particular, we show that one can learn the action of a unitary on entangled states having trained only product states. Since product states can be prepared using only single-qubit gates, this advances the prospects of learning quantum dynamics on near term quantum hardware, and further opens up new methods for both the classical and quantum compilation of quantum circuits.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (51)
  1. M.Ā Schuld, I.Ā Sinayskiy,Ā andĀ F.Ā Petruccione,Ā An introduction to quantum machine learning,Ā Contemporary PhysicsĀ 56,Ā 172 (2015).
  2. M.Ā SchuldĀ andĀ F.Ā Petruccione,Ā Machine Learning with Quantum ComputersĀ (Springer,Ā 2021).
  3. Y.Ā Liu, S.Ā Arunachalam,Ā andĀ K.Ā Temme,Ā A rigorous and robust quantum speed-up in supervised machine learning,Ā Nature PhysicsĀ ,Ā 1 (2021).
  4. D.Ā Aharonov, J.Ā Cotler,Ā andĀ X.-L.Ā Qi,Ā Quantum algorithmic measurement,Ā Nature CommunicationsĀ 13,Ā 1 (2022).
  5. H.-Y.Ā Huang, R.Ā Kueng,Ā andĀ J.Ā Preskill,Ā Information-theoretic bounds on quantum advantage in machine learning,Ā Phys. Rev. Lett.Ā 126,Ā 190505 (2021a).
  6. J.Ā Cotler, H.-Y.Ā Huang,Ā andĀ J.Ā R.Ā McClean,Ā Revisiting dequantization and quantum advantage in learning tasks,Ā arXiv preprint arXiv:2112.00811Ā  (2021).
  7. M.Ā C.Ā CaroĀ andĀ I.Ā Datta,Ā Pseudo-dimension of quantum circuits,Ā Quantum Machine IntelligenceĀ 2,Ā 14 (2020).
  8. L.Ā Banchi, J.Ā Pereira,Ā andĀ S.Ā Pirandola,Ā Generalization in quantum machine learning: A quantum information standpoint,Ā PRX QuantumĀ 2,Ā 040321 (2021).
  9. C.Ā M.Ā Popescu,Ā Learning bounds for quantum circuits in the agnostic setting,Ā Quantum Information ProcessingĀ 20,Ā 1 (2021).
  10. H.Ā Cai, Q.Ā Ye,Ā andĀ D.-L.Ā Deng,Ā Sample complexity of learning parametric quantum circuits,Ā Quantum Science and TechnologyĀ 7,Ā 025014 (2022).
  11. T.Ā Volkoff, Z.Ā Holmes,Ā andĀ A.Ā Sornborger,Ā Universal compiling and (no-)free-lunch theorems for continuous-variable quantum learning,Ā PRX QuantumĀ 2,Ā 040327 (2021).
  12. J.Ā Preskill,Ā Quantum computing in the NISQ era and beyond,Ā QuantumĀ 2,Ā 79 (2018).
  13. C.Ā L.Ā Degen, F.Ā Reinhard,Ā andĀ P.Ā Cappellaro,Ā Quantum sensing,Ā Rev. Mod. Phys.Ā 89,Ā 035002 (2017).
  14. T.Ā JonesĀ andĀ S.Ā C.Ā Benjamin,Ā Robust quantum compilation and circuit optimisation via energy minimisation,Ā QuantumĀ 6,Ā 628 (2022).
  15. H.-Y.Ā Hu, S.Ā Choi,Ā andĀ Y.-Z.Ā You,Ā Classical shadow tomography with locally scrambled quantum dynamics,Ā Physical Review ResearchĀ 5,Ā 023027 (2023).
  16. V.Ā N.Ā VapnikĀ andĀ A.Ā Y.Ā Chervonenkis,Ā On the uniform convergence of relative frequencies of events to their probabilities,Ā Th. Prob. App.Ā 16,Ā 264 (1971).
  17. L.Ā G.Ā Valiant,Ā A theory of the learnable,Ā Communications of the ACMĀ 27,Ā 1134 (1984).
  18. D.Ā GottesmanĀ andĀ I.Ā Chuang,Ā Quantum digital signatures,Ā arXiv preprint quant-ph/0105032Ā  (2001).
  19. H.Ā Shimodaira,Ā Improving predictive inference under covariate shift by weighting the log-likelihood function,Ā Journal of statistical planning and inferenceĀ 90,Ā 227 (2000).
  20. S.Ā J.Ā PanĀ andĀ Q.Ā Yang,Ā A survey on transfer learning,Ā IEEE Transactions on knowledge and data engineeringĀ 22,Ā 1345 (2009).
  21. Note, we use the terms ā€˜distributionā€™ and ā€˜ensembleā€™ interchangeably.
  22. Here and elsewhere, the ā€œāˆ¼similar-to\simāˆ¼ā€ notation means that the random variable on the left has the distribution on the right as its law. For instance, Uāˆ¼š’°LSsimilar-toš‘ˆsubscriptš’°LSU\sim\mathcal{U}_{\rm LS}italic_U āˆ¼ caligraphic_U start_POSTSUBSCRIPT roman_LS end_POSTSUBSCRIPT means that the random unitary Uš‘ˆUitalic_U is drawn from the distribution š’°LSsubscriptš’°LS\mathcal{U}_{\rm LS}caligraphic_U start_POSTSUBSCRIPT roman_LS end_POSTSUBSCRIPT.
  23. Here and elsewhere, š’°ā¢|0āŸ©āŠ—nš’°superscriptket0tensor-productabsentš‘›\mathcal{U}|0\rangle^{\otimes n}caligraphic_U | 0 āŸ© start_POSTSUPERSCRIPT āŠ— italic_n end_POSTSUPERSCRIPT denotes the ensemble of states generated by drawing unitaries from š’°š’°\mathcal{U}caligraphic_U and applying them to the nš‘›nitalic_n-qubit all-zero state |0āŸ©āŠ—nsuperscriptket0tensor-productabsentš‘›|0\rangle^{\otimes n}| 0 āŸ© start_POSTSUPERSCRIPT āŠ— italic_n end_POSTSUPERSCRIPT.
  24. D.Ā StilckĀ FranƧaĀ andĀ R.Ā Garcia-Patron,Ā Limitations of optimization algorithms on noisy quantum devices,Ā Nature PhysicsĀ 17,Ā 1221 (2021).
  25. H.-Y.Ā Huang, R.Ā Kueng,Ā andĀ J.Ā Preskill,Ā Predicting many properties of a quantum system from very few measurements,Ā Nature PhysicsĀ 16,Ā 1050 (2020).
  26. J.Ā A.Ā NelderĀ andĀ R.Ā Mead,Ā A simplex method for function minimization,Ā The Computer JournalĀ 7,Ā 308 (1965).
  27. E.Ā KnillĀ andĀ R.Ā Laflamme,Ā Power of one bit of quantum information,Ā Physical Review LettersĀ 81,Ā 5672 (1998).
  28. D.Ā A.Ā RobertsĀ andĀ B.Ā Yoshida,Ā Chaos and complexity by design,Ā Journal of High Energy PhysicsĀ 2017,Ā 121 (2017).
  29. S.Ā T.Ā FlammiaĀ andĀ J.Ā J.Ā Wallman,Ā Efficient estimation of pauli channels,Ā ACM Transactions on Quantum ComputingĀ 1,Ā 1 (2020).
  30. R.Ā Harper, S.Ā T.Ā Flammia,Ā andĀ J.Ā J.Ā Wallman,Ā Efficient learning of quantum noise,Ā Nature PhysicsĀ 16,Ā 1184 (2020).
  31. R.Ā Harper, W.Ā Yu,Ā andĀ S.Ā T.Ā Flammia,Ā Fast estimation of sparse quantum noise,Ā PRX QuantumĀ 2,Ā 010322 (2021).
  32. S.Ā T.Ā FlammiaĀ andĀ R.Ā Oā€™Donnell,Ā Pauli error estimation via population recovery,Ā QuantumĀ 5,Ā 549 (2021).
  33. M.Ā C.Ā Caro,Ā Binary classification with classical instances and quantum labels,Ā Quantum Machine IntelligenceĀ 3,Ā 18 (2021).
  34. M.Ā Fanizza, Y.Ā Quek,Ā andĀ M.Ā Rosati,Ā Learning quantum processes without input control,Ā arXiv preprint arXiv:2211.05005Ā  (2022).
  35. H.-Y.Ā Huang, S.Ā T.Ā Flammia,Ā andĀ J.Ā Preskill,Ā Foundations for learning from noisy quantum experiments,Ā arXiv preprint arXiv:2204.13691Ā  (2022b).
  36. H.-Y.Ā Huang, S.Ā Chen,Ā andĀ J.Ā Preskill,Ā Learning to predict arbitrary quantum processes,Ā arXiv preprint arXiv:2210.14894Ā  (2022c).
  37. M.Ā C.Ā Caro,Ā Learning quantum processes and Hamiltonians via the Pauli transfer matrix,Ā arXiv preprint arXiv:2212.04471Ā  (2022).
  38. A.Ā Uvarov, A.Ā Kardashin,Ā andĀ J.Ā D.Ā Biamonte,Ā Machine learning phase transitions with a quantum processor,Ā Physical Review AĀ 102,Ā 012415 (2020).
  39. M.Ā Arjovsky,Ā Out of distribution generalization in machine learning,Ā arXiv preprint arXiv:2103.02667Ā  (2021).
  40. M.Ā A.Ā Nielsen,Ā A simple formula for the average gate fidelity of a quantum dynamical operation,Ā Physics Letters AĀ 303,Ā 249 (2002).
  41. C.Ā B.Ā MendlĀ andĀ M.Ā M.Ā Wolf,Ā Unital quantum channelsā€“convex structure and revivals of Birkhoffā€™s theorem,Ā Communications in Mathematical PhysicsĀ 289,Ā 1057 (2009).
  42. F.Ā G.Ā Brandao, A.Ā W.Ā Harrow,Ā andĀ M.Ā Horodecki,Ā Local random quantum circuits are approximate polynomial-designs,Ā Communications in Mathematical PhysicsĀ 346,Ā 397 (2016).
  43. A.Ā HarrowĀ andĀ S.Ā Mehraban,Ā Approximate unitary tš‘”titalic_t-designs by short random quantum circuits using nearest-neighbor and long-range gates,Ā arXiv preprint arXiv:1809.06957Ā  (2018).
  44. J.Ā Haferkamp,Ā Random quantum circuits are approximate unitary tš‘”titalic_t-designs in depth Oā¢(nā¢t5+oā¢(1))š‘‚š‘›superscriptš‘”5š‘œ1O\left(nt^{5+o(1)}\right)italic_O ( italic_n italic_t start_POSTSUPERSCRIPT 5 + italic_o ( 1 ) end_POSTSUPERSCRIPT ),Ā QuantumĀ 6,Ā 795 (2022).
  45. D.Ā Gross, K.Ā Audenaert,Ā andĀ J.Ā Eisert,Ā Evenly distributed unitaries: On the structure of unitary designs,Ā Journal of mathematical physicsĀ 48,Ā 052104 (2007).
  46. R.Ā KuengĀ andĀ D.Ā Gross,Ā Qubit stabilizer states are complex projective 3-designs,Ā arXiv preprint arXiv:1510.02767Ā  (2015).
  47. Z.Ā Webb,Ā The clifford group forms a unitary 3-design,Ā Quantum Information and ComputationĀ 16,Ā 1379 (2016).
  48. H.Ā Zhu,Ā Multiqubit clifford groups are unitary 3-designs,Ā Physical Review AĀ 96,Ā 062336 (2017).
  49. F.Ā VatanĀ andĀ C.Ā Williams,Ā Optimal quantum circuits for general two-qubit gates,Ā Physical Review AĀ 69,Ā 032315 (2004).
  50. V.Ā V.Ā Shende, I.Ā L.Ā Markov,Ā andĀ S.Ā S.Ā Bullock,Ā Minimal universal two-qubit controlled-not-based circuits,Ā Physical Review AĀ 69,Ā 062321 (2004).
  51. R.Ā R.Ā Tucci,Ā An Introduction to Cartanā€™s KAK Decomposition for QC Programmers,Ā arXiv:quant-ph/0507171Ā  (2005).
Citations (72)
View on Semantic Scholar

Summary

No one has generated a summary of this paper yet.

Sign Up to Summarize

Paper to Video (Beta)

No one has generated a video about this paper yet.

Sign Up to Generate All Videos Subscribe on YouTube

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Sign Up to Generate

Open Problems

We haven't generated a list of open problems mentioned in this paper yet.

Generate Now

Continue Learning

We haven't generated follow-up questions for this paper yet.

Generate Now

Collections

Sign up for free to add this paper to one or more collections.

Sign Up

Tweets

Sign up for free to view the 1 tweet with 0 likes about this paper.

Sign Up for Free