Federated Learning Data Validation via External Corroboration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current federated learning systems do not validate local data on client devices prior to training, which can lead to inaccurate training data being used, compromising the accuracy of the global model.
Innovation Solution
The system corroborates client device data by comparing it to external datasets both locally and externally, ensuring the accuracy of the training data and improving the model's accuracy by preventing the use of inaccurate data for training.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If federated learning systems train models using local data on client devices without validation, then data confidentiality is maintained, but the accuracy of the global model deteriorates due to use of inaccurate training data
Solution Approach 1:
The patent introduces an intermediary validation system that acts as a mediator between the federated learning process and external data sources. This validation system verifies local data accuracy by comparing against external datasets without requiring direct access to sensitive training data, thus maintaining confidentiality while improving model reliability through validated training data
Solution Approach 2:
The patent applies preliminary validation of local training data before the federated learning training process begins. By pre-validating data accuracy using external sources and generating validation reports, the system prevents inaccurate data from entering the training pipeline, thereby improving global model accuracy without compromising data confidentiality during the actual training process
2Measurement precision
If external validation data is accessed to corroborate local training data, then the accuracy of training data is improved, but the complexity of the system increases due to additional validation mechanisms
Solution Approach 1:
The patent extracts only the essential validation functionality from the complex federated learning system by creating a separate, dedicated validation system. This extracted validation component interacts with external data sources independently, performing accuracy verification without requiring integration into the core training architecture, thus improving measurement precision while minimizing added complexity
Solution Approach 2:
The patent uses copying by creating validation copies or representations of local training data that can be compared against external datasets. Instead of directly accessing or transmitting sensitive training data, the system generates copies or metadata representations for validation purposes, enabling accurate verification while maintaining data confidentiality and reducing system complexity
Data Source
AI summary
Systems and methods for federated learning including validation of training data on client devices before training models based on the training data. In some aspects, the system accesses local data to be used for training a local model on a client device. The system generates event metadata corresponding to events in the local data. The event metadata is based on validation data received from an external server. The validation data is inaccessible to a central server. The system determines to exclude a portion of the local data that is not validated by the event metadata and generates validated local data. The system trains the local model based on the validated local data. The system processes, using a data de-identification function, the event metadata to generate de-identified event metadata. The system transmits the local model and the de-identified event metadata to the central server.