Class-Based Parameter Prediction for Combinatorial Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for solving combinatorial optimization problems face inefficiencies due to inappropriate parameter settings, leading to suboptimal performance in reaching better solutions.
Innovation Solution
A data processing system that classifies problem instances into classes based on feature vectors, determines class centroid coordinates, extracts instance groups along axial directions, and generates trained models to predict optimal parameter values for each class, using unit vectors to correct generic values for improved solving performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If generic parameter values are used for solving combinatorial optimization problems, then the solving process is simple and fast, but the solving performance is suboptimal due to inappropriate parameter settings
Solution Approach 1:
The patent changes parameter values based on the classification of problem instances. Different parameter sets are selected according to the instance class, allowing optimal parameter configuration for each problem type while maintaining efficient solving processes
Solution Approach 2:
The patent applies different parameter settings to different local groups of instances based on their classification. Each instance class receives customized parameter values tailored to its specific characteristics, rather than using a single universal parameter set
2Reliability
If individual parameter optimization is performed for each instance, then the solving performance is maximized, but the computational overhead and time consumption increase significantly
Solution Approach 1:
The patent performs parameter optimization in advance by classifying instances and determining optimal parameters for each class before actual solving. This preliminary classification and parameter assignment eliminates the need for time-consuming individual optimization during the solving process
Solution Approach 2:
The patent creates universal parameter sets for each instance class that can be applied to multiple instances within that class. This multi-functional approach allows a single parameter set to serve many instances, reducing redundant optimization work
Data Source
AI summary
A processing unit determines each axial direction of a class coordinate system having an origin corresponding to a class centroid, based on feature vectors of instances in the same class, obtains, for each second instance group corresponding to the axial directions, a set of second generic values for parameters, generates unit vectors representing correction directions for a set of first generic values, based a on first feature vector corresponding to the second instance group, the class centroid coordinates, and the sets of first and second generic values, creates a trained model that receives an instance in the class and outputs coordinates in the class coordinate system, and calculates, for a new instance classified into the class, values for the parameters for solving the new instance, based on first coordinates of the new instance obtained using the trained model, the set of first generic values, and the unit vectors.