Search Point Determination Using Gaussian Process Area Sizing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current parameter search methods, such as those based on Bayesian optimization, face challenges in optimizing parameters within a time limit, as they often result in biased search points and continuous searching of low-confidence points near closing time, making it difficult to complete parameter optimization efficiently.
Innovation Solution
A search point determining method that calculates search prediction time and confidence interval upper limits using a Gaussian process, generates areas in parameter space for each search candidate point, and determines search points based on the size of these areas, allowing for a balance between wide searches and reusing evaluation results according to available time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If Bayesian optimization is used to perform parameter search by predicting solution location, then processing amount is suppressed, but search points become biased and optimization cannot be completed efficiently within time limit
Solution Approach 1:
The patent pre-calculates and stores evaluation results at multiple search points before the actual optimization process. When a search point is needed, the pre-computed results are reused directly, avoiding repeated calculations and enabling rapid parameter optimization within time constraints
Solution Approach 2:
The patent dynamically adjusts the search strategy by switching between using pre-stored evaluation results and performing new calculations based on remaining time and search progress. This dynamic adaptation allows efficient utilization of available time while maintaining search effectiveness
2Loss of time
If search results are reused near completed search points, then calculation time is shortened, but search coverage becomes limited and biased
Solution Approach 1:
The patent segments the parameter space into multiple regions and maintains evaluation results for each region. This segmentation allows the system to reuse results within specific regions while still being able to explore other regions, balancing search coverage with calculation efficiency
Solution Approach 2:
The patent changes the parameters of the search process by adjusting which pre-stored results are reused and which new points are explored, based on the current search state and time constraints. This flexible parameter adjustment maintains both search coverage and efficiency
3Reliability
If continuous searching is performed near closing time, then search thoroughness is maintained, but time efficiency decreases and optimization cannot be completed
Solution Approach 1:
The patent implements feedback mechanisms that continuously monitor the search progress and remaining time. Based on this feedback, the system adjusts its strategy by reusing pre-stored evaluation results when time is limited, ensuring that optimization can be completed while maintaining acceptable thoroughness
Data Source
AI summary
A search point determining method in an estimation process of a function, executed by a processor included in a search point determining apparatus, the method includes, calculating a search prediction time and a confidence interval upper limit obtained by using a Gaussian process for the function in each search candidate point from a past search result of the function, generating an area in a parameter space for each search candidate point by using a position of a search point close to the relevant search candidate point in a past search result, a search prediction time corresponding to each search candidate point, and a confidence interval upper limit corresponding to each search candidate point, and determining a search point based on a size of the area in a plurality of parameter spaces.