Historic Metadata Updating With Thresholded ML Screening
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems face inefficiencies in updating metadata for historic data due to the need for accurate detection of objects of interest, which often require high-capacity machine learning models, leading to resource wastage and inaccuracy when applied to data with low probability of containing the object.
Innovation Solution
A method that employs a low-capacity machine learning model to assess the probability of historic data containing an object of interest, switching to a high-capacity model for confirmation and metadata update when the probability exceeds a threshold, thereby conserving resources while maintaining accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-capacity machine learning models are used to detect objects of interest in historic data, then detection accuracy is improved, but computational resource consumption increases
Solution Approach 1:
The detection process is segmented into two stages: first, a low-capacity model performs initial screening to identify potential candidates with above-threshold probability, and second, a high-capacity model is applied only to these candidates for final confirmation. This segmentation resolves the contradiction by limiting high-capacity model usage to only necessary cases.
Solution Approach 2:
A preliminary filtering action is performed using the low-capacity model before applying the high-capacity model. The low-capacity model pre-identifies which historic data items warrant further analysis, thereby preventing unnecessary consumption of computational resources on items unlikely to contain objects of interest.
2Use of energy by moving object
If low-capacity machine learning models are used to analyze historic data, then computational resources are conserved, but detection accuracy decreases
Solution Approach 1:
The low-capacity model serves as an intermediary that bridges the gap between resource constraints and accuracy requirements. It performs initial assessment and filters candidates, allowing the high-capacity model to focus only on promising cases, thus achieving both resource efficiency and accuracy.
Solution Approach 2:
Instead of applying the high-capacity model to all historic data (excessive action), the system applies it only to a subset of data identified as high-probability candidates by the low-capacity model (partial action), achieving sufficient accuracy while conserving resources.
3Reliability
If high-capacity machine learning models are applied to all historic data, then comprehensive detection is achieved, but processing time increases
Solution Approach 1:
The processing workflow is segmented into rapid initial filtering followed by detailed analysis only of relevant items. This segmentation maintains comprehensive detection for high-probability cases while avoiding time-consuming processing of low-probability items.
Solution Approach 2:
Preliminary filtering by the low-capacity model quickly eliminates low-probability candidates, reducing the volume of data that requires time-consuming high-capacity model analysis, thus overall reducing processing time while maintaining detection reliability.
4Productivity
If low-capacity machine learning models are used for all data, then processing speed is improved, but detection reliability decreases
Solution Approach 1:
The system dynamically adjusts model capacity based on the specific data item being processed. The low-capacity model handles the majority of cases efficiently, while the high-capacity model is dynamically activated for cases requiring higher reliability, optimizing both speed and reliability adaptively.
Solution Approach 2:
The low-capacity model acts as an intermediary that handles routine cases quickly while identifying cases that require the more reliable but slower high-capacity model, thus achieving overall improved processing speed without sacrificing detection reliability for critical cases.
Data Source
AI summary
A system for updating metadata associated with historic data. The system includes an electronic computing device. The electronic computing device is configured to receive historic data and metadata associated with the historic data and, using a low capacity machine learning model, analyze the metadata associated with the historic data to determine a probability that the historic data includes the object of interest. The electronic computing device is also configured to compare the probability to a predetermined threshold. The electronic computing device is further configured to use a high capacity machine learning model to analyze the historic data to determine whether the historic data includes the object of interest and update the metadata of the historic data based on whether the historic data includes the object of interest to generate updated metadata, when the probability is greater than or equal to the predetermined threshold.