Scalable Federated Learning for Clients with Different Input Image Sizes and Numbers of Output Categories
This addresses a practical issue in federated learning for real-world applications with diverse client data, but it is incremental as it builds on existing federated learning methods.
The paper tackles the problem of federated learning with clients having varying input image sizes and output categories by proposing ScalableFL, which adjusts local model depths and widths accordingly, and demonstrates its effectiveness in heterogeneous settings for image classification and object detection tasks.
Federated learning is a privacy-preserving training method which consists of training from a plurality of clients but without sharing their confidential data. However, previous work on federated learning do not explore suitable neural network architectures for clients with different input images sizes and different numbers of output categories. In this paper, we propose an effective federated learning method named ScalableFL, where the depths and widths of the local models for each client are adjusted according to the clients' input image size and the numbers of output categories. In addition, we provide a new bound for the generalization gap of federated learning. In particular, this bound helps to explain the effectiveness of our scalable neural network approach. We demonstrate the effectiveness of ScalableFL in several heterogeneous client settings for both image classification and object detection tasks.