Ising Model Optimization for Robust Solution Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing optimization algorithms struggle to provide robust solutions that can withstand variations in real-world conditions, leading to reduced effectiveness when deviations occur from the exact solution.
Innovation Solution
A method using a binary variable sampling technique, specifically the FMQA, which involves creating an Ising model in the QUBO format, varying recommended input variables, calculating evaluation values, and updating the training data set with these values to achieve a robust solution while minimizing calculation cost.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing optimization algorithms are used to find the optimal solution, then the evaluation value is maximized, but the solution becomes sensitive to variations and loses robustness in real-world conditions
Solution Approach 1:
The patent applies preliminary action by proactively generating multiple candidate solutions and pre-evaluating their performance before actual deployment. The system creates a diverse set of solutions through controlled variations and stores them in a database, allowing the system to prepare for potential variations in advance rather than reacting after deviations occur
Solution Approach 2:
The patent implements feedback mechanisms by continuously monitoring the performance of generated solutions and using this information to refine future solution generation. The system evaluates solution performance, compares it against expected outcomes, and adjusts the generation parameters accordingly, creating a closed-loop optimization process that learns from actual performance data
2Reliability
If multiple candidate solutions are generated and stored to improve robustness, then the solution reliability increases, but the data management complexity and calculation cost increase
Solution Approach 1:
The patent applies local quality by differentiating the treatment of different solution characteristics. Rather than uniformly managing all solution data, the system identifies and prioritizes specific attributes and parameters that are most critical for robustness, applying enhanced management and evaluation only to those local aspects that matter most for solution reliability
Solution Approach 2:
The patent implements partial action by generating and storing only the necessary number of candidate solutions required to achieve adequate robustness, rather than exhaustively generating all possible solutions. The system determines an optimal balance point where sufficient solution diversity is achieved without excessive computational resources being consumed
3Measurement precision
If the training data set is continuously updated with new evaluation data, then the model accuracy improves, but the calculation time and processing cost increase
Solution Approach 1:
The patent applies periodic action by implementing batched updates to the training data set rather than continuous real-time updates. The system accumulates evaluation data over discrete time periods and performs model retraining at scheduled intervals, allowing the model to be updated periodically with new information without requiring constant computational resources
Solution Approach 2:
The patent implements preliminary action by pre-processing and structuring evaluation data in advance before it is needed for model updates. The system prepares data in standardized formats and organizes it for efficient ingestion, allowing rapid integration of new evaluation results without requiring extensive processing time at the moment of update
Data Source
AI summary
A non-transitory computer-readable recording medium storing an arithmetic program for causing a computer to repeatedly execute a process includes creating an Ising model based on a training data set that includes pieces of training data in which an input variable and an evaluation value are associated, searching for a recommended input variable based on the Ising model, giving a variation to the recommended input variable and calculating the evaluation value for the recommended input variable after the variation, and adding the recommended input variable and the evaluation value to the training data set when the recommended input variable does not exist in the training data, and updating a statistical value of the evaluation value of the training data and the evaluation value for the recommended input variable after the variation as the evaluation value of the training data when the recommended input variable exists in the training data.