Training Data Deletion Control in Sensor Data Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data management systems face challenges in efficiently selecting and deleting training data for model training, leading to increased data storage burdens and risks of data leakage, as they do not effectively differentiate between data suitable for training and abnormal data.
Innovation Solution
A data management system that acquires and stores measurement data from sensors, selectively transmits training data to a learning unit for model training, and deletes the data after confirmation of successful training, using a prohibition list and permission list to manage data deletion based on predetermined criteria.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If training data is deleted after transmission to learning unit, then data storage volume is reduced, but risk of data leakage increases if deletion is not confirmed
Solution Approach 1:
The data management system implements a feedback mechanism where the learning unit transmits training completion information back to confirm successful training before data deletion occurs. This feedback loop ensures that data is only deleted after verified successful training, eliminating data leakage risk while achieving storage reduction.
Solution Approach 2:
The system performs preliminary verification by receiving training completion information from the learning unit before executing data deletion. This preliminary action ensures that data deletion only occurs under confirmed safe conditions, resolving the contradiction between storage reduction and leakage prevention.
2Adaptability or versatility
If all measurement data is retained for potential training use, then data availability is maintained, but data storage burden increases
Solution Approach 1:
The system discards training data from storage after successful training completion is confirmed, while maintaining the capability to retrieve and re-train using the same data if needed. This approach reduces storage burden while preserving data availability for future training needs.
Solution Approach 2:
The system extracts only the necessary training data from the measurement data for training purposes, transmits it to the learning unit, and then removes it from storage. This extraction approach maintains data availability during the training process while reducing overall storage burden by removing used training data.
3Productivity
If data deletion is performed without confirmation of training completion, then system operation speed is improved, but manufacturing precision of training model deteriorates due to premature data removal
Solution Approach 1:
The system uses feedback from the learning unit regarding training completion status to control the data deletion timing. This feedback mechanism ensures data is retained during the entire training process and only deleted after confirmed completion, maintaining model quality while enabling automated efficient management.
Solution Approach 2:
The training completion information acts as an intermediary signal that mediates between the learning unit and the data management system. This intermediary mechanism coordinates the deletion action with training completion, ensuring neither premature deletion nor unnecessary data retention occurs.
Data Source
AI summary
A data management system is provided. The data management system includes: a data acquisition unit for acquiring measurement data on a measurement target; a data storage unit for storing the measurement data; a data transmission unit for transmitting training data in the measurement data, used for training a model, to a learning unit for training the model; and a data deletion unit for deleting the training data from the stored measurement data.