Leverage Variational Graph Representation for Model Poisoning on Federated Learning

This article 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.

I. Introduction

Federated learning (FL) has attracted significant attention recently, and emerged as a distributed deep learning paradigm. With FL, each user device trains its local model with its private data to generate local updates sent to the edge server without sharing the device’s private data. The edge server then aggregates the local updates to train a global model, which is sent back to the user devices for the next round of FL training. Based on FL, individual data privacy is protected as no private data is shared [1].