Iterative Feature Selection for Computational Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Iterative modeling processes face significant computational challenges when dealing with high-dimensional datasets, leading to excessively long computation times due to the need to search all possible model component combinations, which is impractical for solving complex problems with large datasets.
Innovation Solution
The method involves generating models using a subset of dataset components, calculating model-attribute and utility metrics, and computing weighted utility metrics to eliminate or retain model components, thereby reducing the number of components available for modeling processes, specifically using techniques like genetic programming to identify and prune unnecessary components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all possible model component combinations are searched in iterative modeling processes, then model accuracy and reliability are improved, but computation time increases exponentially
Solution Approach 1:
The patent applies preliminary action by pre-calculating utility metrics for all model components before the main iterative modeling process. This preliminary evaluation ranks components based on their individual utility, allowing the system to prioritize and select only the most promising components for inclusion in model combinations, thereby avoiding exhaustive search of all possible combinations while still achieving high model accuracy
Solution Approach 2:
The patent extracts and removes low-utility model components from the search space by calculating utility metrics and eliminating components below a certain threshold. This extraction process reduces the dimensionality of the problem by focusing only on the most relevant components, significantly decreasing computation time while maintaining model reliability
2Loss of time
If the universe of algorithm components is reduced to decrease computational requirements, then computation time is decreased, but model accuracy may be compromised
Solution Approach 1:
The patent changes the parameter of component selection by introducing utility metrics as a new criterion for evaluating model components. Instead of using traditional methods of component selection, the system calculates utility metrics that quantify the contribution of each component, allowing for more intelligent and accurate reduction of the component universe while preserving model accuracy
Solution Approach 2:
The patent implements feedback by using utility metrics to guide the iterative modeling process. The utility metric calculations provide feedback information about component importance, which is then used to adjust the search space and component selection in subsequent iterations, ensuring that only the most relevant components are considered while maintaining model accuracy
Data Source
AI summary
Feature selection methods and processes that facilitate reduction of model components available for iterative modeling. It has been discovered that methods of eliminating model components that do not meaningfully contribute to a solution can be preliminarily discovered and discarded, thereby dramatically decreasing computational requirements in iterative programming techniques. This development unlocks the ability of iterative modeling to be used to solve complex problems that, in the past, would have required computation time on orders of magnitude too great to be useful.