Ising Machine Data Input Apparatus for Automated QUBO Conversion

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The manual conversion of optimization problems from their usual form to Quadratic Unconstrained Binary Optimization (QUBO) form is difficult, time-consuming, and requires specialized expertise, often resulting in complex and inefficient formulations.

Innovation Solution

A processor-based system that converts optimization problems into suitable formats for Ising machines by creating datasets of input-output pairs, fitting mathematical expressions, and assessing them against quality metrics to ensure efficient QUBO or higher-order formulations, allowing for fully automatic or semi-automatic conversion.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If manual conversion to QUBO form is performed by human experts, then the optimization problem can be solved on an Ising Machine, but the process is difficult, time-consuming and requires specialized expertise

Engineering Contradiction:
Improveconversion speedVSAvoidformulation complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent replaces the manual mechanical process of QUBO formulation with an automated computer-based system. The system includes a converter that automatically converts optimization problems to QUBO form, and a trainer that generates training data to improve the converter's accuracy over time, eliminating the need for manual expert intervention

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system enables self-service automation where the converter automatically performs QUBO formulation without requiring human experts. The trainer component further enhances this by autonomously generating training data and improving the converter's performance through machine learning, making the system progressively more autonomous

Inventive Principle:
Principle #25Self-service

2Reliability

If manual QUBO formulation is performed, then the problem can be translated for Ising Machine input, but it results in complex and inefficient formulations

Engineering Contradiction:
Improvesolution correctnessVSAvoidconversion efficiency
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent implements feedback through the trainer component that evaluates the converter's output and generates training data based on the difference between manual expert formulations and automated converter output. This feedback loop continuously improves the converter's accuracy and efficiency, ensuring both correctness and productivity

Inventive Principle:
Principle #23Feedback

3Ease of operation

If automated conversion is implemented, then the need for specialized expertise is reduced, but ensuring the correctness of the global optimum becomes challenging

Engineering Contradiction:
Improveuser accessibilityVSAvoidoptimization accuracy
Core Design Contradiction:
Ease of operationVSMeasurement precision

Solution Approach 1:

The patent performs preliminary action by training the converter extensively before deployment. The trainer generates large amounts of training data and iteratively improves the converter's performance, ensuring it achieves high accuracy before being used for actual optimization problems, thus maintaining measurement precision while improving ease of operation

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS12001812B2Ising machine data input apparatus and method of inputting data into an Ising machine
Publication Date: 2024.06.04 FUJITSU LTD
  • US12001812B2 patent drawing
  • US12001812B2 patent drawing
  • US12001812B2 patent drawing

AI summary

Apparatus and method of inputting data into an Ising machine. The apparatus may comprise at least one processor to carry out a conversion process to convert an input expression, in a format unsuitable for inputting into an Ising machine, to a suitable format, including creating a dataset of input-output data pairs on the basis of the input expression, deriving a mathematical expression by fitting a first dataset subset to coefficients of an exemplary mathematical expression in the suitable format, and using a second dataset subset, different from the first, to assess whether the derived expression meets a preset quality metric. The derived expression is input to the Ising machine when it is assessed as meeting the metric. The conversion process may be repeated using a different input expression when the derived expression is assessed as failing to meet the metric.