Area Generation Device Using Gradient Search for Event Distribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods fail to accurately extract areas with dense event occurrences from a two-dimensional plane, either by strictly limiting shape and size or resulting in overly complex shapes that do not naturally represent the event distribution.
Innovation Solution
A computer-readable storage medium and device that generates an adjacency matrix from divided areas, uses an evaluation function combining distribution data and adjacency relations, and applies a gradient method search to determine optimal areas with dense event occurrences, allowing for flexible shape evaluation and reducing complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If strict shape and size limitations are imposed on extracted areas, then the extraction process becomes simpler and more controlled, but the fitting accuracy to actual event distributions deteriorates
Solution Approach 1:
The invention changes the parameter of area shape from fixed geometric constraints to flexible boundary definitions based on event distribution data. The boundary is dynamically adjusted to fit the actual spatial pattern of events, transforming the extraction from a rigid geometric process to an adaptive statistical process that maintains simplicity while improving accuracy.
Solution Approach 2:
The extracted area boundaries are made dynamic rather than static. Instead of using predetermined geometric shapes, the boundaries continuously adapt to the underlying event distribution patterns in the data, allowing the extraction process to respond to the actual spatial characteristics of the events being analyzed.
2Measurement precision
If no shape restrictions are imposed on extracted areas, then the fitting accuracy to event distributions improves, but the resulting shapes become overly complex and lose interpretability
Solution Approach 1:
The invention applies different qualities to different parts of the extracted area. The boundary follows the event distribution closely in regions where events are dense, while maintaining smoother transitions in regions with lower event density. This local adaptation allows high fitting accuracy where needed while preventing overall shape complexity from becoming excessive.
Solution Approach 2:
Instead of creating entirely new complex shapes, the invention copies and adapts the underlying event distribution pattern itself to define the area boundary. The boundary essentially traces the spatial footprint of the events, creating a shape that naturally fits the data without requiring complex geometric constructions.
3Ease of operation
If conventional extraction methods are used, then the processing approach remains simple and intuitive, but the ability to accurately represent natural event distributions deteriorates
Solution Approach 1:
The invention replaces mechanical geometric construction methods with statistical and computational approaches. Instead of manually defining or rigidly constraining area shapes, the system uses algorithms to automatically determine boundaries based on event distribution patterns, substituting computational processing for mechanical geometric operations.
Data Source
AI summary
A storage medium storing a program that causes a processor to execute for acquiring distribution data indicating a distribution of spots in an area to be searched, and area data indicating positions of divided areas obtained by dividing the area to be searched; generating an adjacency matrix indicating an adjacency relation between the divided areas from the area data; generating an evaluation function for evaluating selection of the divided areas using: a variable indicating selection of consecutive divided areas, the distribution data, and the adjacency matrix; calculating a gradient of a value of the evaluation function from a current value of the variable; executing a gradient method search for updating the value of the variable using the calculated gradient; and determining selection of the divided areas as an optimal area for event occurrence analysis based on a result of the gradient method search.