Crossfire: An Elastic Defense Framework for Graph Neural Networks Under Bit Flip Attacks

TL;DR

This paper introduces Crossfire, a novel, retraining-free defense framework for Graph Neural Networks against Bit Flip Attacks, effectively restoring network integrity and improving prediction quality with minimal overhead.

Contribution

The paper proposes Crossfire, the first hybrid, verifiably restoring defense for GNNs against BFAs, combining hashing, honeypots, and bit-level correction without retraining or labeled data.

Findings

Crossfire outperforms existing defenses with a 21.8% higher probability of restoring GNNs to pre-attack state.

It improves post-repair prediction quality by 10.85%.

Effective against up to 55 bit flips across six benchmark datasets.

Abstract

Bit Flip Attacks (BFAs) are a well-established class of adversarial attacks, originally developed for Convolutional Neural Networks within the computer vision domain. Most recently, these attacks have been extended to target Graph Neural Networks (GNNs), revealing significant vulnerabilities. This new development naturally raises questions about the best strategies to defend GNNs against BFAs, a challenge for which no solutions currently exist. Given the applications of GNNs in critical fields, any defense mechanism must not only maintain network performance, but also verifiably restore the network to its pre-attack state. Verifiably restoring the network to its pre-attack state also eliminates the need for costly evaluations on test data to ensure network quality. We offer first insights into the effectiveness of existing honeypot- and hashing-based defenses against BFAs adapted from the computer vision domain to GNNs, and characterize the shortcomings of these approaches. To overcome their limitations, we propose Crossfire, a hybrid approach that exploits weight sparsity and combines hashing and honeypots with bit-level correction of out-of-distribution weight elements to restore network integrity. Crossfire is retraining-free and does not require labeled data. Averaged over 2,160 experiments on six benchmark datasets, Crossfire offers a 21.8% higher probability than its competitors of reconstructing a GNN attacked by a BFA to its pre-attack state. These experiments cover up to 55 bit flips from various attacks. Moreover, it improves post-repair prediction quality by 10.85%. Computational and storage overheads are negligible compared to the inherent complexity of even the simplest GNNs.