Labeling Chaos to Learning Harmony: Federated Learning with Noisy Labels
This work addresses label noise in federated learning, a critical issue for improving model accuracy in distributed systems with noisy user-generated data, though it is incremental as it builds on existing FL methods.
The paper tackles the problem of label noise in federated learning, where existing methods perform poorly due to data scarcity and varying noise levels across clients, and proposes FedLN, a framework that estimates noise levels and corrects or mitigates noisy samples, achieving a 22% average improvement on synthetic noisy datasets and a 4.8% increase on real-world datasets.
Federated Learning (FL) is a distributed machine learning paradigm that enables learning models from decentralized private datasets, where the labeling effort is entrusted to the clients. While most existing FL approaches assume high-quality labels are readily available on users' devices; in reality, label noise can naturally occur in FL and is closely related to clients' characteristics. Due to scarcity of available data and significant label noise variations among clients in FL, existing state-of-the-art centralized approaches exhibit unsatisfactory performance, while prior FL studies rely on excessive on-device computational schemes or additional clean data available on server. Here, we propose FedLN, a framework to deal with label noise across different FL training stages; namely, FL initialization, on-device model training, and server model aggregation, able to accommodate the diverse computational capabilities of devices in a FL system. Specifically, FedLN computes per-client noise-level estimation in a single federated round and improves the models' performance by either correcting or mitigating the effect of noisy samples. Our evaluation on various publicly available vision and audio datasets demonstrate a 22% improvement on average compared to other existing methods for a label noise level of 60%. We further validate the efficiency of FedLN in human-annotated real-world noisy datasets and report a 4.8% increase on average in models' recognition performance, highlighting that~\method~can be useful for improving FL services provided to everyday users.