FB-NLL: A Feature-Based Approach to Tackle Noisy Labels in Personalized Federated Learning

2026-04-21

AI Summary

This work addresses the challenge of label noise and low-quality data in personalized federated learning, which often leads to biased user clustering and degraded model performance. The authors propose a label-free, one-shot clustering framework based on feature analysis: by examining the spectral structure of local feature covariances and employing a geometry-aware subspace similarity metric, the method achieves robust user grouping. Furthermore, it introduces a direction-aligned feature consistency mechanism that corrects intra-cluster label noise without requiring estimation of the noise transition matrix. Notably, the approach is training-dynamics-agnostic and model-agnostic, demonstrating superior performance across diverse datasets and noise configurations. It consistently outperforms existing methods in both average accuracy and stability.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Search and Optimization: Local Search
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Responsible Web: Data and user privacy-enhancing technologies for the Web
• Web Mining and Content Analysis: Normalization, clustering, classification, and summarization of Web text

Abstract

Personalized Federated Learning (PFL) aims to learn multiple task-specific models rather than a single global model across heterogeneous data distributions. Existing PFL approaches typically rely on iterative optimization-such as model update trajectories-to cluster users that need to accomplish the same tasks together. However, these learning-dynamics-based methods are inherently vulnerable to low-quality data and noisy labels, as corrupted updates distort clustering decisions and degrade personalization performance. To tackle this, we propose FB-NLL, a feature-centric framework that decouples user clustering from iterative training dynamics. By exploiting the intrinsic heterogeneity of local feature spaces, FB-NLL characterizes each user through the spectral structure of the covariances of their feature representations and leverages subspace similarity to identify task-consistent user groupings. This geometry-aware clustering is label-agnostic and is performed in a one-shot manner prior to training, significantly reducing communication overhead and computational costs compared to iterative baselines. Complementing this, we introduce a feature-consistency-based detection and correction strategy to address noisy labels within clusters. By leveraging directional alignment in the learned feature space and assigning labels based on class-specific feature subspaces, our method mitigates corrupted supervision without requiring estimation of stochastic noise transition matrices. In addition, FB-NLL is model-independent and integrates seamlessly with existing noise-robust training techniques. Extensive experiments across diverse datasets and noise regimes demonstrate that our framework consistently outperforms state-of-the-art baselines in terms of average accuracy and performance stability.

Problem

• Personalized Federated Learning
• Noisy Labels
• User Clustering
• Label Noise
• Data Heterogeneity

Innovation

• Feature-Based Clustering
• Noisy Label Learning
• Personalized Federated Learning
• Subspace Similarity
• Label-Agnostic Clustering