Tracking Data Validation Using Feature Constraints and Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing tracking data often contains errors that limit the accuracy and reliability of subsequent processing or visualization tasks, especially when data is combined from multiple sources.
Innovation Solution
An apparatus and method for validating tracking data by receiving samples, generating features, performing error detection using constraints, and validating the data based on the detection results, utilizing machine learning models to normalize and identify errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If tracking data is obtained from multiple data providers, then the quantity and coverage of tracking data is improved, but the reliability and accuracy deteriorate due to errors in the tracking data
Solution Approach 1:
The patent introduces an intermediary validation system that acts as a mediator between multiple data providers and the tracking data processing pipeline. This validation apparatus receives tracking data from multiple providers, performs error detection using machine learning models and constraint-based validation, and filters out erroneous data before it reaches subsequent processing stages. The intermediary validation layer enables the system to maintain high data quantity from multiple sources while ensuring reliability through automated error detection and validation rules.
2Reliability
If error detection and validation processes are implemented, then the reliability of tracking data is improved, but the device complexity increases due to additional processing steps
Solution Approach 1:
The validation system is segmented into distinct functional modules: error detection module using machine learning models, constraint-based validation module, and data filtering module. Each module performs a specific validation function independently, allowing the system to achieve high reliability through specialized processing while managing complexity through modular architecture. The segmentation enables parallel processing of different validation techniques without requiring a monolithic complex system.
Solution Approach 2:
The validation apparatus employs self-service mechanisms where machine learning models automatically learn error patterns from training data and continuously improve detection accuracy without manual intervention. The system performs self-validation by comparing tracking data against learned constraints and statistical models, enabling automated reliability improvement without requiring complex manual validation processes or extensive human oversight.
3Measurement precision
If machine learning models are used for error detection, then the measurement precision of tracking data is improved, but the use of energy increases due to computational requirements
Solution Approach 1:
Machine learning models for error detection are trained in advance using preliminary action, where extensive training data is processed offline to build pre-trained models that capture error patterns and constraints. During actual tracking data validation, these pre-trained models perform rapid inference with minimal computational energy requirements compared to real-time training. The preliminary training phase separates the energy-intensive learning process from the low-energy validation process, enabling high measurement precision during operation without excessive energy consumption.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
In accordance with the present disclosure, an apparatus for validation of tracking data from a data provider is provided, the apparatus comprising circuitry configured to: receive a first sample of tracking data from a data provider, the first sample of tracking data comprising information related to the location of at least one person within a physical environment at a first instance of time; generate information related to one or more features of the at least one person using the sample of tracking data; perform error detection using the information related to the one or more features of the at least one person which has been generated and one or more constraints on features of a person in the physical environment; and validate the first sample of tracking data in accordance with a result of the error detection.