Data Classification Using Salient-Part Selection for State Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data classification methods lack the ability to effectively utilize additional knowledge about the data to improve classification accuracy, particularly in technical systems where automated state determination is crucial.
Innovation Solution
A computer-implemented method and apparatus that determines a first classification using a model, identifies the most salient parts of the data, and supplements the data with these parts to achieve a second classification, which can improve the accuracy of state determination in technical systems by using either the same model or a pre-trained model independently of the system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data are classified using a standard model without additional processing, then the classification process is simple and fast, but the classification accuracy is limited
Solution Approach 1:
The patent applies preliminary action by identifying and selecting the most salient parts of the data before performing the final classification. The system calculates salience measures for different data parts, selects the top-k most relevant parts, and uses these pre-selected parts as enhanced input for classification, thereby improving accuracy without requiring complete reprocessing of all data
Solution Approach 2:
The patent extracts the most salient parts of the data from the complete dataset based on calculated salience measures. By taking out only the most relevant portions of the data (top-k salient parts) and using them for classification, the system improves classification accuracy while avoiding the complexity of processing entire datasets, thus resolving the contradiction between precision and complexity
2Measurement precision
If all parts of the data are used for classification, then complete information is available, but the processing time and computational resources increase
Solution Approach 1:
The system extracts only the most salient parts of the data (top-k parts with highest salience measures) for classification instead of processing all data parts. This extraction approach maintains classification accuracy by focusing on the most relevant information while significantly reducing processing time and computational resource requirements
Solution Approach 2:
The patent applies partial action by using only a subset of the most important data parts (top-k salient parts) rather than all available data. This partial processing approach achieves sufficient classification accuracy without the time and resource costs of complete data processing, effectively resolving the time-accuracy trade-off
Data Source
AI summary
An apparatus and computer-implemented method for classifying data. A first classification is determined as a function of the data using a first model. A measure for a contribution of the relevant part to the classification is in each case determined for parts of the data to be classified. The data are supplemented by at least one of the parts, for which, as a function of the measures determined for the parts, it is determined that the at least one of the parts to be classified contributes more to the first classification than other parts to be classified. A second classification is determined as a function of the data supplemented by the at least one of the parts using the first model or using a second model.