Barren plateaus in quantum neural network training landscapes

Abstract

Many experimental proposals for noisy intermediate scale quantum devices involve training a parameterized quantum circuit with a classical optimization loop. Such hybrid quantum-classical algorithms are popular for applications in quantum simulation, optimization, and machine learning. Due to its simplicity and hardware efficiency, random circuits are often proposed as initial guesses for exploring the space of quantum states. We show that the exponential dimension of Hilbert space and the gradient estimation complexity make this choice unsuitable for hybrid quantum-classical algorithms run on more than a few qubits. Specifically, we show that for a wide class of reasonable parameterized quantum circuits, the probability that the gradient along any reasonable direction is non-zero to some fixed precision is exponentially small as a function of the number of qubits. We argue that this is related to the 2-design characteristic of random circuits, and that solutions to this problem must be studied.

Full text

Its licence does not allow us to show the full text. Read it at PubMed Central:

References 51

1. 1.
   Quantum Computing in the NISQ era and beyond · 2018

   Not checked
2. 2.

   Peruzzo A, et al. A variational eigenvalue solver on a photonic quantum processor. Nat. Commun. 2014;5:1.

   Not in our library10.1038/ncomms5213

3. 3.

   McClean JR, Romero J, Babbush R, Aspuru-Guzik A. The theory of variational hybrid quantum-classical algorithms. New J. Phys. 2016;18:023023.

   Not in our library10.1088/1367-2630/18/2/023023

4. 4.

   Yung MH, et al. From transistor to trappedion computers for quantum chemistry. Sci. Rep. 2014;4:9.

   Not in our library10.1038/srep03589

5. 5.

   Farhi, E. Goldstone, J. & Gutmann, S. A quantum approximate optimization algorithm. Preprint at https://arxiv.org/abs/1411.4028 (2014).

   Not in our library

6. 6.

   Johnson, P. D., Romero, J., Olson, J., Cao, Y. & Aspuru-Guzik, A. QVECTOR: an algorithm for device-tailored quantum error correction. Preprint at https://arxiv.org/abs/1711.02249 (2017).

   Not in our library

7. 7.

   Cao, Y., Giacomo Guerreschi, G. & Aspuru-Guzik, A. Quantum neuron: an elementary building block for machine learning on quantum computers. Preprint at https://arxiv.org/abs/1711.11240 (2017).

   Not in our library

8. 8.

   Hempel C, et al. Quantum chemistry calculations on a trappedion quantum simulator. Phys. Rev. X. 2018;8:031022.

   Not in our library

9. 9.

   O’alley PJJ, et al. Scalable quantum simulation of molecular energies. Phys. Rev. X. 2016;6:31007.

   Not in our library

10. 10.

    McClean JR, Schwartz ME, Carter J, de Jong WA. Hybrid quantum-classical hierarchy for mitigation of decoherence and determination of excited states. Phys. Rev. A. 2017;95:42308.

    Not in our library10.1103/physreva.95.042308

11. 11.

    Wecker D, Hastings MB, Troyer M. Progress towards practical quantum variational algorithms. Phys. Rev. A. 2015;92:42303.

    Not in our library10.1103/physreva.92.042303

12. 12.

    Shen Y, et al. Quantum implementation of unitary coupled cluster for simulating molecular electronic structure. Phys. Rev. A. 2017;95:020501(R).

    Not in our library10.1103/physreva.95.020501

13. 13.

    Kandala A, et al. Hardware-efficient quantum optimizer for small molecules and quantum magnets. Nature. 2017;549:242.

    Not in our library10.1038/nature23879

14. 14.

    Colless JI, et al. Computation of molecular spectra on a quantum processor with an error-resilient algorithm. Phys. Rev. X. 2018;8:011021.

    Not in our library

15. 15.

    Santagati R, et al. Witnessing eigenstates for quantum simulation of hamiltonian spectra. Sci. Adv. 2018;4:1.

    Not in our library10.1126/sciadv.aap9646

16. 16.

    Dumitrescu EF, et al. Cloud quantum computing of an atomic nucleus. Phys. Rev. Lett. 2018;120:210501.

    Not in our library10.1103/physrevlett.120.210501

17. 17.

    Wecker D, Hastings MB, Troyer M. Training a quantum optimizer. Phys. Rev. A. 2016;94:022309.

    Not in our library10.1103/physreva.94.022309

18. 18.

    Wang Z, Hadfield S, Jiang Z, Rieffel EG. Quantum approximate optimization algorithm for maxcut: a fermionic view. Phys. Rev. A. 2018;97:022304.

    Not in our library10.1103/physreva.97.022304

19. 19.

    Moll N, et al. Quantum optimization using variational algorithms on nearterm quantum devices. Quantum Sci. Technol. 2018;3:030503.

    Not in our library10.1088/2058-9565/aab822

20. 20.

    Otterbach, J. S. et al. Unsupervised machine learning on a hybrid quantum computer. Preprint at https://arxiv.org/abs/1712.05771v1 (2017).

    Not in our library

21. 21.

    Romero J, Olson JP, Aspuru-Guzik A. Quantum autoencoders for efficient compression of quantum data. Quantum Sci. Technol. 2017;2:045001.

    Not in our library10.1088/2058-9565/aa8072

22. 22.

    Biamonte J, et al. Quantum machine learning. Nature. 2017;549:195.

    Not in our library10.1038/nature23474

23. 23.

    Farhi,E. & Neven, H. Classification with quantum neural networks on near term processors. Preprint at https://arxiv.org/abs/1802.06002 (2018).

    Not in our library

24. 24.
    Deep learning · 2015

    No notice
25. 25.

    McClean JR, Babbush R, Love PJ, Aspuru-Guzik A. Exploiting locality in quantum computation for quantum chemistry. J. Phys. Chem. Lett. 2014;5:4368.

    Not in our library10.1021/jz501649m

26. 26.

    Romero Jonathan, Babbush Ryan, McClean Jarrod R, Hempel Cornelius, Love Peter J, Aspuru-Guzik Alán. Strategies for quantum computing molecular energies using the unitary coupled cluster ansatz. Quantum Science and Technology. 2018;4(1):014008.

    Not in our library10.1088/2058-9565/aad3e4

27. 27.

    Babbush R, et al. Low-depth quantum simulation of materials. Phys. Rev. X. 2018;8:011044.

    Not in our library

28. 28.

    Rubin NC, Babbush R, McClean J. Application of fermionic marginal constraints to hybrid quantum algorithms. New J. Phys. 2018;20:053020.

    Not in our library10.1088/1367-2630/aab919

29. 29.

    Kivlichan ID, et al. Quantum simulation of electronic structure with linear depth and connectivity. Phys. Rev. Lett. 2018;120:110501.

    Not in our library10.1103/physrevlett.120.110501

30. 30.

    Farhi, E., Goldstone, J., Gutmann, S. & Neven, H. Quantum algorithms for fixed qubit architectures. Preprint at http://arxiv.org/abs/1703.06199 (2017).

    Not in our library

31. 31.

    Boixo S, et al. Characterizing quantum supremacy in nearterm devices. Nat. Phys. 2018;14:595.

    Not in our library10.1038/s41567-018-0124-x

32. 32.

    Bradley, D. M. Learning in Modular Systems (Carnegie Mellon University, Pittsburgh, 2010).

    Not in our library

33. 33.

    Glorot, X. & Bengio, Y. Understanding the difficulty of training deep feedforward neural networks. In Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, PMLR. AISTATS. (eds.) Yee Whye Teh, Mike Titterington. 249–256 (2010)

    Not in our library

34. 34.

    Shalev-Shwartz, S., Shamir, O., & Shammah, S. Failures of gradient-based deep learning. Preprint at https://arxiv.org/abs/1703.07950 (2017).

    Not in our library

35. 35.

    Ioffe S. & Szegedy, C. Batch normalization: accelerating deep network training by reducing internal covariate shift. In Proceedings of the 32nd International Conference on Machine Learning, PMLR. ICML. (eds.) Francis Bach, David Blei. 448–456 (2015)

    Not in our library

36. 36.

    Knill E, Ortiz G, Somma RD. Optimal quantum measurements of expectation values of observables. Phys. Rev. A. 2007;75:012328.

    Not in our library10.1103/physreva.75.012328

37. 37.

    Popescu S, Short AJ, Winter A. Entanglement and the foundations of statistical mechanics. Nat. Phys. 2006;2:754.

    Not in our library10.1038/nphys444

38. 38.

    Bremner MJ, Mora C, Winter A. Are random pure states useful for quantum computation? Phys. Rev. Lett. 2009;102:190502.

    Not in our library10.1103/physrevlett.102.190502

39. 39.

    Gross D, Flammia ST, Eisert J. Most quantum states are too entangled to be useful as computational resources. Phys. Rev. Lett. 2009;102:190501.

    Not in our library10.1103/physrevlett.102.190501

40. 40.

    Ledoux, M. The Concentration of Measure Phenomenon (American Mathematical Society, Providence, 2005).

    Not in our library

41. 41.

    Guerreschi, G. G. & Smelyanskiy, M. Practical optimization for hybrid quantum-classical algorithms. Preprint at https://arxiv.org/abs/1701.01450 (2017).

    Not in our library

42. 42.

    Renes JM, Blume-Kohout R, Scott AJ, Caves CM. Symmetric informationally complete quantum measurements. J. Math. Phys. 2004;45:2171.

    Not in our library10.1063/1.1737053

43. 43.

    Dankert C, Cleve R, Emerson J, Livine E. Exact and approximate unitary 2-designs and their application to fidelity estimation. Phys. Rev. A. 2009;80:012304.

    Not in our library10.1103/physreva.80.012304

44. 44.

    Harrow AW, Low RA. Random quantum circuits are approximate 2-designs. Commun. Math. Phys. 2009;291:257.

    Not in our library10.1007/s00220-009-0873-6

45. 45.

    Ipsen, J. R. Products of independent Gaussian random matrices. Preprint at https://arxiv.org/abs/1510.06128 (2015).

    Not in our library

46. 46.

    Roberts DA, Yoshida B. Chaos and complexity by design. J. High. Energy Phys. 2017;2017:121.

    Not in our library10.1007/jhep04(2017)121

47. 47.

    Hochreiter, S., Bengio, Y., Frasconi, P. & Schmidhuber, J. Gradient flow in recurrent nets: the difficulty of learning long-term dependencies. A Field Guide to Dynamical Recurrent Neural Networks, Chapter 14, (eds.) S. C. Kremer and J. F. Kolen. (IEEE Press Piscataway, NJ 2001) https://www.amazon.com/Field-Guide-Dynamical-Recurrent-Networks/dp/0780353692

    Not in our library

48. 48.

    Hinton GE, Osindero S, Teh YW. A fast learning algorithm for deep belief nets. Neural Comput. 2006;18:1527.

    Not in our library10.1162/neco.2006.18.7.1527

49. 49.

    He, K., Zhang, X., Ren, S. & J. Sun, J. Deep residual learning for image recognition. In Proceedings of the IEEE conference on Computer Vision and Pattern Recognition. (Eds.) Raman, B., Kumar, S., Roy, P.P., Sen, D., Las Vegas, NV, United States, 770–778 (2016).

    Not in our library

50. 50.

    Bengio, Y., Lamblin, P., Popovici, D. & Larochelle, H. Greedy layer-wise training of deep networks. Advances in Neural Information Processing Systems 19 (NIPS'06). (eds.) B. Schölkopf, J. C. Platt, T. Hoffman. 53–160 (MIT Press, Canada 2007).

    Not in our library

51. 51.

    Puchałla Z, Miszczak JA. Symbolic integration with respect to the haar measure on the unitary groups. Bull. Pol. Acad. Sci. Tech. Sci. 2017;65:21.

    Not in our library

Cited by papers we hold 6

• Quantum Embedding Search for Quantum Machine Learning · 2022
  Not checked
• Quantum algorithms from fluctuation theorems: Thermal-state preparation · 2022
  Not checked
• Learning from Physics Experiments with Quantum Computers: Applications in Muon Spectroscopy · 2021
  Not checked
• Expressibility of the alternating layered ansatz for quantum computation · 2021
  Not checked
• Machine Learning: Quantum vs Classical · 2020
  Not checked
• Decoding quantum errors with subspace expansions · 2020
  Not checked

Reads

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.