Leverage Variational Graph Representation For Model Poisoning on Federated Learning

TL;DR

This paper introduces a novel model poisoning attack on federated learning that uses a variational graph autoencoder to craft malicious models without access to training data, evading detection and significantly degrading model accuracy.

Contribution

It presents the VGAE-MP attack, a new data-untethered method leveraging graph structures to generate undetectable malicious models in federated learning.

Findings

VGAE-MP causes a significant accuracy drop in federated learning.

Existing defenses are ineffective against VGAE-MP.

The attack exploits graph correlations among benign models.

Abstract

This paper puts forth a new training data-untethered model poisoning (MP) attack on federated learning (FL). The new MP attack extends an adversarial variational graph autoencoder (VGAE) to create malicious local models based solely on the benign local models overheard without any access to the training data of FL. Such an advancement leads to the VGAE-MP attack that is not only efficacious but also remains elusive to detection. VGAE-MP attack extracts graph structural correlations among the benign local models and the training data features, adversarially regenerates the graph structure, and generates malicious local models using the adversarial graph structure and benign models' features. Moreover, a new attacking algorithm is presented to train the malicious local models using VGAE and sub-gradient descent, while enabling an optimal selection of the benign local models for training the VGAE. Experiments demonstrate a gradual drop in FL accuracy under the proposed VGAE-MP attack and the ineffectiveness of existing defense mechanisms in detecting the attack, posing a severe threat to FL.