Partially Local Federated Learning for Privacy-Preserving Model Updates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing federated learning frameworks compromise user privacy by transmitting sensitive user-level data updates outside the user device, which can lead to the recovery of private information.
Innovation Solution
Implement a partially local federated learning framework that uses local training techniques to maintain and update a set of local parameters on the user device, employing reconstruction and client update algorithms to generate parameter updates without transferring raw data or sensitive information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If federated learning frameworks transmit user-level data updates outside the user device, then model training accuracy is improved, but user privacy is compromised
Solution Approach 1:
The patent segments the model parameters into different components that can be updated locally on user devices. Instead of transmitting complete model updates or raw data, only specific parameter updates are transmitted to the server, reducing the risk of privacy leakage while maintaining training effectiveness.
Solution Approach 2:
The patent introduces an intermediate representation of model parameters that acts as a mediator between local user data and the central server. This intermediate form allows information exchange for model improvement without directly exposing sensitive user-level data, thus protecting privacy while enabling accurate training.
2Object-affected harmful factors
If local parameters are maintained and updated on user devices, then user privacy is protected, but model training efficiency may be reduced
Solution Approach 1:
The patent implements partial local training where only specific parameters are updated locally on user devices rather than complete model retraining. This partial action approach maintains privacy protection while reducing computational overhead and improving training efficiency compared to full local training.
Solution Approach 2:
The patent performs preliminary updates on selected model parameters locally on user devices before transmitting them to the server. This preliminary local processing reduces the amount of data that needs to be transmitted and processed centrally, thereby improving overall training efficiency while maintaining privacy.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a machine learning model having a set of local model parameters and a set of global model parameters under a partially local federated learning framework. One of the methods include maintaining local data and data defining the local model parameters; receiving data defining current values of the global model parameters; determining, based on the local data, the local model parameters, and the current values of the global model parameters, current values of the local model parameters; determining, based on the local data, the current values of the local model parameters, and the current values of the global model parameters, updated values of the global model parameters; generating, based on the updated values of the global model parameters, parameter update data defining an update to the global model parameters; and transmitting the parameter update data.