AI/ML Data Quality Evaluation via Bitmap Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data collection mechanisms for artificial intelligence (AI)/machine learning (ML) in cellular networks face challenges in efficiently evaluating and selecting high-quality data samples for training and validation purposes, particularly in 5G positioning systems.
Innovation Solution
The proposed solution involves an apparatus that receives bitmap information from second apparatuses indicating the locations of data samples collected for AI/ML services. This apparatus determines the quality level of the data samples based on the bitmap information and compares it to a target quality level to select appropriate data samples for inclusion in a dataset. The selected data samples are then requested to be reported to a target entity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If all collected data samples are transmitted for AI/ML training, then the dataset completeness is improved, but the signaling overhead increases
Solution Approach 1:
The patent applies partial action by transmitting only a subset of data samples (those meeting quality criteria) rather than all collected samples. The network device receives bitmap information indicating locations of collected samples, evaluates quality levels for each location, and selectively requests transmission of only high-quality samples, thus reducing signaling overhead while maintaining dataset usefulness.
2Quantity of substance
If data samples with varying quality levels are included in the dataset, then the data quantity is improved, but the dataset quality decreases
Solution Approach 1:
The patent implements local quality by evaluating and selecting data samples based on location-specific quality levels. The network device receives bitmap information where each bit corresponds to a specific location, determines quality levels for samples at each location, and selectively includes only samples meeting quality thresholds in the final dataset. This ensures high dataset quality while maintaining sufficient data quantity across diverse locations.
3Area of stationary object
If comprehensive data collection is performed across all locations, then the data coverage is improved, but the data evaluation complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the data collection area into discrete locations represented in bitmap information. Each location is independently evaluated for data sample quality. This segmented approach allows comprehensive spatial coverage while simplifying evaluation, as the network device can process quality assessments location-by-location rather than dealing with the entire dataset as a single complex evaluation problem.
Data Source
AI summary
Example embodiments direct to data evaluation for AI/ML. A method comprises: at a first apparatus, receiving, from at least one second apparatus, bitmap information indicating respective locations of data samples collected by the at least one second apparatus, the at least one second apparatus being configured for data collection for an artificial intelligence (AI)/machine learning (ML) service; determining, based at least in part on the bitmap information, a quality level of data samples collected by the at least one second apparatus; selecting data samples to be included in a dataset for the AI/ML service by comparing the quality level with a target quality level for the dataset; and transmitting, to the at least one second apparatus, a request to report the selected data samples to be included in the dataset to a target entity.