Wanlun Ma (Swinburne University of Technology), Derui Wang (CSIRO’s Data61), Ruoxi Sun (The University of Adelaide & CSIRO's Data61), Minhui Xue (CSIRO's Data61), Sheng Wen (Swinburne University of Technology), Yang Xiang (Digital Research & Innovation Capability Platform, Swinburne University of Technology)

Deep Neural Networks (DNNs) are susceptible to backdoor attacks during training. The model corrupted in this way functions normally, but when triggered by certain patterns in the input, produces a predefined target label. Existing defenses usually rely on the assumption of the universal backdoor setting in which poisoned samples share the same uniform trigger. However,
recent advanced backdoor attacks show that this assumption is no longer valid in dynamic backdoors where the triggers vary from input to input, thereby defeating the existing defenses.

In this work, we propose a novel technique, Beatrix (backdoor detection via Gram matrix). Beatrix utilizes Gram matrix to capture not only the feature correlations but also the appropriately high-order information of the representations. By learning class-conditional statistics from activation patterns of normal samples, Beatrix can identify poisoned samples by capturing the anomalies in activation patterns. To further improve the performance in identifying target labels, Beatrix leverages kernel-based testing without making any prior assumptions on representation distribution. We demonstrate the effectiveness of our method through extensive evaluation and comparison with state-of-the-art defensive techniques. The experimental results show that our approach achieves an F1 score of 91.1% in detecting dynamic backdoors, while the state of the art can only reach 36.9%.

View More Papers

A Case Study on Fuzzing Satellite Firmware

Tobias Scharnowski, Felix Buchmann, Simon Woerner, Thorsten Holz Presenter: Tobias Scharnowski

Read More

Brokenwire : Wireless Disruption of CCS Electric Vehicle Charging

Sebastian Köhler (University of Oxford), Richard Baker (University of Oxford), Martin Strohmeier (armasuisse Science + Technology), Ivan Martinovic (University of...

Read More

“I didn't click”: What users say when reporting phishing

Nikolas Pilavakis, Adam Jenkins, Nadin Kokciyan, Kami Vaniea (University of Edinburgh)

Read More

Un-Rocking Drones: Foundations of Acoustic Injection Attacks and Recovery...

Jinseob Jeong (KAIST, Agency for Defense Development), Dongkwan Kim (Samsung SDS), Joonha Jang (KAIST), Juhwan Noh (KAIST), Changhun Song (KAIST),...

Read More