Federated Learning Node Model Update Resists Discrimination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Federated learning lacks a solution to address discrimination propagation, which occurs due to differences in data distribution across interacting nodes, leading to biased model performance and degradation.
Innovation Solution
A method for updating node models that involves calculating a distribution weighted aggregation model, a regularization term, and a class balanced complementary term using a cross-domain feature generator to mitigate discrimination propagation by aligning node models with their specific data distributions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If personalized federated learning is used to improve overall performance, then model performance improvement is enhanced, but data discrimination and performance variance propagate across nodes
Solution Approach 1:
The patent introduces a discriminator as an intermediary component that mediates between the personalized model and the global model. The discriminator detects and blocks discriminatory information propagation while allowing beneficial performance improvements to pass through, thus resolving the contradiction between overall performance improvement and discrimination propagation.
Solution Approach 2:
The patent implements a feedback mechanism where the discriminator provides feedback signals to both the personalized model and the global model. This feedback loop enables the system to continuously adjust and reduce discrimination propagation while maintaining performance improvement, by reinforcing non-discriminative features and suppressing biased patterns.
2Adaptability or versatility
If many idiosyncratic data models are introduced to handle local data distribution, then local adaptability is improved, but bias transfer and discrimination propagation increase
Solution Approach 1:
The patent employs the global model as a counterweight to the idiosyncratic local models. The discriminator ensures that while local models adapt to their specific data distributions, they are counterbalanced by the global model's broader perspective, preventing excessive bias transfer and discrimination propagation across the federated learning system.
Data Source
AI summary
Disclosed is a method for updating a node model that resists discrimination propagation in federated learning. The method includes: obtaining a node model corresponding to a data node; calculating a mean value of the distribution of class features and a quantity ratio corresponding to training data of the data node, calculating a distribution weighted aggregation model based on the node model, the mean value of the distribution of class features and the quantity ratio; calculating a regularization term corresponding to the data node based on the node model and the distribution weighted aggregation model; calculating a variance of the distribution of the class features corresponding to the data node, calculating a class balanced complementary term by using a cross-domain feature generator; and updating the node model based on the distribution weighted aggregation model, the regularization term, and the class balanced complementary term.