Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Addressing Class Imbalance in Federated Learning

About

Federated learning (FL) is a promising approach for training decentralized data located on local client devices while improving efficiency and privacy. However, the distribution and quantity of the training data on the clients' side may lead to significant challenges such as class imbalance and non-IID (non-independent and identically distributed) data, which could greatly impact the performance of the common model. While much effort has been devoted to helping FL models converge when encountering non-IID data, the imbalance issue has not been sufficiently addressed. In particular, as FL training is executed by exchanging gradients in an encrypted form, the training data is not completely observable to either clients or servers, and previous methods for class imbalance do not perform well for FL. Therefore, it is crucial to design new methods for detecting class imbalance in FL and mitigating its impact. In this work, we propose a monitoring scheme that can infer the composition of training data for each FL round, and design a new loss function -- \textbf{Ratio Loss} to mitigate the impact of the imbalance. Our experiments demonstrate the importance of acknowledging class imbalance and taking measures as early as possible in FL training, and the effectiveness of our method in mitigating the impact. Our method is shown to significantly outperform previous methods, while maintaining client privacy.

Lixu Wang, Shichao Xu, Xiao Wang, Qi Zhu• 2020

Related benchmarks

TaskDatasetResultRank
Image ClassificationFashionMNIST (test)--
461
Image ClassificationF-MNIST (test)
Accuracy94
173
Image ClassificationCIFAR-10-LT (IF 50)
Top-1 Accuracy64.77
88
Image ClassificationCIFAR-100 LT (IF=50)
Top-1 Acc36.88
82
Image ClassificationCIFAR-10-LT IF 100
Top-1 Accuracy59.75
78
Image ClassificationCIFAR-100 LT (IF=100)
Top-1 Acc32.95
55
Long-Tailed Image ClassificationImageNet LT
Accuracy (Many)36.33
51
Image ClassificationCIFAR-10-LT IF 10
Top-1 Accuracy78.14
46
Image ClassificationCIFAR10 LT-100
Clean Accuracy22.66
29
Image ClassificationCIFAR-100 LT IF=10
Top-1 Accuracy46.79
25
Showing 10 of 17 rows

Other info

Follow for update