Systems and methods to generate samples for machine learning using quantum computing

Patent metadata

Field	Value
Publication number	US-12229632-B2
Application number	US-202017030576-A
Country	US
Kind code	B2
Filing date	Sep 24, 2020
Priority date	Mar 7, 2016
Publication date	Feb 18, 2025
Grant date	Feb 18, 2025

Abstract

A hybrid computer comprising a quantum processor can be operated to perform a scalable comparison of high-entropy samplers. Performing a scalable comparison of high-entropy samplers can include comparing entropy and KL divergence of post-processed samplers. A hybrid computer comprising a quantum processor generates samples for machine learning. The quantum processor is trained by matching data statistics to statistics of the quantum processor. The quantum processor is tuned to match moments of the data.

First claim

The invention claimed is:

1. A method of operation of a computational system comprising at least one processor, the method comprising:
receiving information defining a target distribution comprising a Boltzmann distribution and a sampling process by the at least one processor;
receiving a plurality of samples using the sampling process by the at least one processor;
generating a sampling distribution based on the plurality of samples by the at least one processor;
generating a post-processed distribution from the plurality of samples by the at least one processor, wherein generating the post-processed distribution includes:
determining a partition defining a first and a second subset of variables by a graphical model defined by the Boltzmann distribution, wherein the partition between the first and the second subset of variables is selected such that a conditional probability distribution of the second subset of variables can be marginalized and an induced sub-graph caused by the partition is defined with respect to the Boltzmann distribution; and
expressing the post-processed distribution in a mixture model comprising the sampling distribution with respect to the first subset of variables and an analytic form with respect to the second subset of variables, suitable for conditional resampling on the first and the second subset of variables;

evaluating a Kullback-Leibler (KL) divergence from the target distribution to the post-processed distribution by the at least one processor; and
comparing the sampling distribution to the target distribution based at least in part on the KL divergence by the at least one processor.

2. The method of claim 1 wherein receiving the plurality of samples using the sampling process by the at least one processor includes receiving the plurality of samples from a quantum processor.

3. The method of claim 2 wherein determining the partition defining the first and second subset of variables by the graphical model further comprises determining the partition defining the first and second subset of variables by the graphical model defined by a hardware graph of the quantum processor.

4. The method of claim 1 , further comprising:
generating, by a quantum processor, the plurality of samples using the sampling process, and wherein receiving the plurality of samples using the sampling process by the at least one processor includes receiving the plurality of samples by a digital processor from the analog-quantum processor.

5. The method of claim 4 wherein generating by the quantum processor the plurality of samples using the sampling process includes generating by the quantum processor the plurality of samples using quantum annealing.

6. The method of claim 1 wherein generating the sampling distribution based on the plurality of samples by the at least one processor includes generating an empirical distribution.

7. A computing system comprising at least one processor for comparing a sampling distribution and a target distribution, the computing system operable to:
receive information defining the target distribution comprising a Boltzmann distribution and a sampling process associated with the sampling distribution by the at least one processor;
receive a plurality of samples by the at least one processor, wherein the plurality of samples are generated using the sampling process;
generate the sampling distribution based on the plurality of samples by the at least one processor;
generate a post-processed distribution from the plurality of samples by the at least one processor, wherein to generate the post-processed distribution by the at least one processor, the at least one processor:
determines a partition defining a first and a second subset of variables by a graphical model defined by the Boltzmann distribution, wherein the partition between the first and the second subset of variables is selected such that a conditional probability distribution of the second subset of variables can be marginalized and an induced sub-graph caused by the partition is defined with respect to the Boltzmann distribution;
expresses the post-processed distribution in a mixture model comprising the sampling distribution with respect to the first subset of variables and an analytic form with respect to the second subset of variables, suitable for conditional resampling on the first and the second subset of variables;
evaluates a Kullback-Leibler (KL) divergence from the target distribution to the post-processed distribution; and
compares the sampling distribution to the target distribution based at least in part on the KL divergence.

8. The computing system of claim 7 wherein the at least one processor comprises a digital processor and a quantum processor.

9. The computing system of claim 8 wherein the computing system is operable to receive the plurality of samples by the digital processor, wherein the plurality of samples are generated by the quantum processor using the sampling process.

10. The computing system of claim 9 wherein the plurality of samples are generated by the quantum processor using quantum annealing.

11. The computing system of claim 7 wherein the sampling distribution is an empirical distribution.

12. The computing system of claim 7 wherein the graphical model is further defined by a hardware graph of a quantum processor.

13. A hybrid computing system comprising at least one digital processor and a quantum processor for comparing a sampling distribution and a target distribution, the target distribution comprising a Boltzmann distribution, the hybrid computing system operable to:
receive information defining the target distribution and a sampling process by the at least one digital processor;
generate a plurality of samples by the quantum processor using the sampling process;
receive the plurality of samples by the at least one digital processor;
generate the sampling distribution based on the plurality of samples by the at least one digital processor;
generate a post-processed distribution from the plurality of samples by the at least one digital processor, wherein to generate the post-processed distribution by the at least one digital processor, the at least one digital processor:
determines a partition defining a first and a second subset of variables by a graphical model defined by the Boltzmann distribution, wherein the partition between the first and the second subset of variables is selected such that a conditional probability distribution of the second subset of variables can be marginalized and an induced sub-graph caused by the partition is defined with respect to the Boltzmann distribution;
expresses the post-processed distribution in a mixture model comprising the sampling distribution with respect to the first subset of variables and an analytic form with respect to the second subset of variables, suitable for conditional resampling on the first and the second subset of variables;
evaluates a Kullback-Leibler (KL) divergence from the target distribution to the post-processed distribution by the at least one digital processor; and
compares the sampling distribution to the target distribution based at least in part on the KL divergence by the at least one digital processor.

14. The hybrid computing system of claim 13 wherein the sampling distribution is an empirical distribution.

15. The hybrid computing system of claim 13 wherein the plurality of samples are generated by the quantum processor using quantum annealing.

16. The hybrid computing system of claim 13 wherein the graphical model is further defined by a hardware graph of

Assignees

• D Wave Systems Inc

Inventors

• Macready William G
• Hamze Firas
• Chudak Fabian A
• Ranjbar Mani
• Raymond Jack R
• Rolfe Jason T

Classifications

• G06N10/00 Primary

  Quantum computing, i.e. information processing based on quantum-mechanical phenomena · CPC title

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06F18/2415

  based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate · CPC title

• G06F2111/10

  Numerical modelling · CPC title

• G06N20/00

  Machine learning · CPC title