Federated Learning Decision Tree Model Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Federated learning methods face challenges in constructing effective models while ensuring data privacy and reducing communication overhead, as the iterative process of adjusting model parameters between participants and servers consumes significant resources, making it difficult to protect privacy and reduce communication costs simultaneously.
Innovation Solution
A federated learning method that determines candidate features from a training dataset, constructs decision tree models based on these features, selects a second decision tree model based on prediction results, and transmits it to another device for fusion, thereby reducing the need for multiple transmissions and minimizing communication overhead while maintaining data privacy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If iterative adjustment of model parameters is performed between participant and federated server, then data privacy and security are protected, but communication overhead increases significantly
Solution Approach 1:
The patent segments the model parameters into two parts: local data remains with the participant while only encrypted model parameters are transmitted to the federated server. This segmentation allows privacy protection while reducing the volume of data that needs to be communicated iteratively.
Solution Approach 2:
The patent uses encrypted copies of model parameters that can be transmitted without exposing the actual data. The federated server works with these encrypted copies, adjusting them locally and returning modified copies, thereby protecting privacy while enabling collaboration.
2Manufacturing precision
If multiple transmissions of model parameters are performed to achieve standard convergence, then model accuracy is improved, but communication consumption increases
Solution Approach 1:
The patent performs preliminary encryption of model parameters before transmission, and the federated server performs preliminary adjustments locally. This preliminary action reduces the need for multiple back-and-forth transmissions, as the server can process and adjust parameters in advance without requiring continuous data exchange.
Solution Approach 2:
The patent extracts only the necessary model parameters for adjustment from the full dataset, transmitting only these encrypted parameter subsets to the federated server. This extraction minimizes communication consumption while maintaining the ability to achieve model convergence through localized adjustments.
Data Source
AI summary
A federated learning method includes: determining at least one candidate feature from data features corresponding to a training data-set, the candidate feature corresponding to at least two decision trends in a decision tree model; obtaining n first decision tree models by taking the at least one candidate feature as a model construction foundation, value of n corresponding to number of the at least one candidate feature; determining at least one second decision tree model from the n first decision tree models based on prediction results of the n first decision tree models on training data in the training data-set; and transmitting the second decision tree model to a second computing device, the second computing device being configured to fuse at least two decision tree models that comprise the second decision tree model to obtain a federated learning model.