Adiabatic training for Variational Quantum Algorithms

Acosta, Ernesto; Gutierrez, Carlos Cano; Botella, Guillermo; Campos, Roberto

Adiabatic training for Variational Quantum Algorithms

Oct 24, 2024

12 pages

e-Print:

2410.18618 [quant-ph]

View in:

ADS Abstract Service

pdf

reference search0 citations

Citations per year

0 Citations

Abstract: (arXiv)

This paper presents a new hybrid Quantum Machine Learning (QML) model composed of three elements: a classical computer in charge of the data preparation and interpretation; a Gate-based Quantum Computer running the Variational Quantum Algorithm (VQA) representing the Quantum Neural Network (QNN); and an adiabatic Quantum Computer where the optimization function is executed to find the best parameters for the VQA. As of the moment of this writing, the majority of QNNs are being trained using gradient-based classical optimizers having to deal with the barren-plateau effect. Some gradient-free classical approaches such as Evolutionary Algorithms have also been proposed to overcome this effect. To the knowledge of the authors, adiabatic quantum models have not been used to train VQAs. The paper compares the results of gradient-based classical algorithms against adiabatic optimizers showing the feasibility of integration for gate-based and adiabatic quantum computing models, opening the door to modern hybrid QML approaches for High Performance Computing.

Note:

12 pages, 6 figures, Euro PAR 2024 EuroQHPC Workshop

References(18)

Figures(6)

[1]

: Quantum Computation and Quantum Information. 10th edn. Cambridge University Press

M. Nielsen
,
I. Chuang

[2]

: Machine Learning with Quantum Computers. 2nd edn

M. Schuld
,
F. Petruccione

[3]

Quantum Computation by Adiabatic Evolution

Edward Farhi
(
- MIT, Cambridge, CTP
)
,
Jeffrey Goldstone
(
- MIT, Cambridge, CTP
)
,
Sam Gutmann
(
- Northeastern U., Math. Dept.
)
,
Michael Sipser
(
- MIT
)

e-Print:
- quant-ph/0001106

[4]

Algorithms for Training Variational Quantum Classifiers

G. Acampora
,
A. Chiatto
,
A. Vitiello
,
A. Comparison of Evolutionary

[5]

Variational quantum algorithms

M. Cerezo
(
- Los Alamos and
- Oak Ridge
)
,
Andrew Arrasmith
(
- Los Alamos and
- Oak Ridge
)
,
Ryan Babbush
(
- Unlisted, US, CA
)
,
Simon C. Benjamin
(
- Oxford U.
)
,
Suguru Endo
(
- Nippon NTT, Kanagawa
)

et al.

- Nature Rev.Phys. 3 (2021) 9, 625-644
•
e-Print:
- 2012.09265
•
DOI:
- 10.1038/s42254-021-00348-9

[6]

: Artificial Intelligence: A Modern Approach

S. Russell
,
P. Norvig

[7]

: Beyond Regression: New Tools for Prediction and Analysis in the Behavioral Sciences. Harvard University

P.J. Werbos

[8]

: An adiabatic method to train binarized artificial neural networks

Y. Zhao
,
J. Xiao

- Sci.Rep. 11 (2021) 19797

[9]

The computer as a physical system: A microscopic quantum mechanical Hamiltonian model of computers as represented by Turing machines

Paul Benioff

- J.Statist.Phys. 22 (1980) 5, 563-591
•
DOI:
- 10.1007/BF01011339

[10]

: On Computable Numbers, with an Application to the Entscheidungsproblem

J. Jung

- Proc.Lond.Math.Soc. 42 (1936) 230-265

[11]

: An Algorithmic Crystal Ball: Forecasts-based on Machine Learning. International Monetary Fund

A. Turing
,
M. Patnam
,
A. Ter-Martirosyan

[12]

: Neural Networks for Survival Prediction in Medicine Using Prognostic Factors: A Review

G. Kantidakis
,
A.D. Hazewinkel
,
M. Fiocco

[13]

: Short term energy consumption forecasting using neural basis expansion analysis for interpretable time series. Nature Scientific Reports 12

A.K. Shaikh
,
A. Nazir
,
I. Khan
,
A.S. Shah

[14]

Quantum recurrent neural networks for sequential learning

et al.

- Neural Networks 166 (2023) 148-161
•
e-Print:
- 2302.03244
•
DOI:
- 10.1016/j.neunet.2023.07.003

[15]

: Input Redundancy for Parameterized Quantum Circuits. Frontiers in Physics 8

F. Gil
,
D. Theis

[16]

Ye, Q., Deng, D

H. Cai

[17]

: Learning representations by backpropagating errors

D. Rumelhart
,
G. Hinton
,
R. Williams

- Nature 232 (1986) 533-536

[18]

: Logical Reversibility of Computation

C.H. Bennett

- IBM J.Res.Dev. 17 (1973) 525-532