Sample Selection Using Multi-Task Autoencoders in Federated Learning with Non-IID Data

TL;DR

This paper introduces novel sample selection techniques using multi-task autoencoders and outlier detection methods to improve federated learning performance on non-IID image data.

Contribution

It proposes a multi-task autoencoder-based approach combined with outlier detection and SVDD loss for effective sample filtering in federated learning.

Findings

Significant accuracy improvements on CIFAR10 and MNIST datasets.

Loss-based sample selection methods outperform baseline approaches.

Feature-based sample selection with SVDD further enhances model accuracy.

Abstract

Federated learning is a machine learning paradigm in which multiple devices collaboratively train a model under the supervision of a central server while ensuring data privacy. However, its performance is often hindered by redundant, malicious, or abnormal samples, leading to model degradation and inefficiency. To overcome these issues, we propose novel sample selection methods for image classification, employing a multitask autoencoder to estimate sample contributions through loss and feature analysis. Our approach incorporates unsupervised outlier detection, using one-class support vector machine (OCSVM), isolation forest (IF), and adaptive loss threshold (AT) methods managed by a central server to filter noisy samples on clients. We also propose a multi-class deep support vector data description (SVDD) loss controlled by a central server to enhance feature-based sample selection. We validate our methods on CIFAR10 and MNIST datasets across varying numbers of clients, non-IID distributions, and noise levels up to 40%. The results show significant accuracy improvements with loss-based sample selection, achieving gains of up to 7.02% on CIFAR10 with OCSVM and 1.83% on MNIST with AT. Additionally, our federated SVDD loss further improves feature-based sample selection, yielding accuracy gains of up to 0.99% on CIFAR10 with OCSVM. These results show the effectiveness of our methods in improving model accuracy across various client counts and noise conditions.