Local K-Similarity Constraint for Federated Learning with Label Noise

TL;DR

This paper introduces a novel regularization method for federated learning with noisy labels, leveraging self-supervised representations to improve robustness without requiring shared architecture or large pretrained models.

Contribution

It proposes a local K-Similarity Constraint that decouples pretrained and classification models, enhancing federated learning robustness against label noise without increasing communication costs.

Findings

Outperforms state-of-the-art federated methods on vision and medical image benchmarks.

Does not require shared architecture between pretrained and classifier models.

Significantly improves model robustness in high-noise, heterogeneous client scenarios.

Abstract

Federated learning on clients with noisy labels is a challenging problem, as such clients can infiltrate the global model, impacting the overall generalizability of the system. Existing methods proposed to handle noisy clients assume that a sufficient number of clients with clean labels are available, which can be leveraged to learn a robust global model while dampening the impact of noisy clients. This assumption fails when a high number of heterogeneous clients contain noisy labels, making the existing approaches ineffective. In such scenarios, it is important to locally regularize the clients before communication with the global model, to ensure the global model isn't corrupted by noisy clients. While pre-trained self-supervised models can be effective for local regularization, existing centralized approaches relying on pretrained initialization are impractical in a federated setting due to the potentially large size of these models, which increases communication costs. In that line, we propose a regularization objective for client models that decouples the pre-trained and classification models by enforcing similarity between close data points within the client. We leverage the representation space of a self-supervised pretrained model to evaluate the closeness among examples. This regularization, when applied with the standard objective function for the downstream task in standard noisy federated settings, significantly improves performance, outperforming existing state-of-the-art federated methods in multiple computer vision and medical image classification benchmarks. Unlike other techniques that rely on self-supervised pretrained initialization, our method does not require the pretrained model and classifier backbone to share the same architecture, making it architecture-agnostic.