ImpNet: Imperceptible and blackbox-undetectable backdoors in compiled neural networks

2022-09-30 · Tim Clifford, Ilia Shumailov, Yiren Zhao, Ross Anderson, Robert Mullins

Early backdoor attacks against machine learning set off an arms race in attack and defence development. Defences have since appeared demonstrating some ability to detect backdoors in models or even remove them. These defences work by inspecting the training data, the model, or the integrity of the training procedure. In this work, we show that backdoors can be added during compilation, circumventing any safeguards in the data preparation and model training stages. The attacker can not only insert existing weight-based backdoors during compilation, but also a new class of weight-independent backdoors, such as ImpNet. These backdoors are impossible to detect during the training or data preparation processes, because they are not yet present. Next, we demonstrate that some backdoors, including ImpNet, can only be reliably detected at the stage where they are inserted and removing them anywhere else presents a significant challenge. We conclude that ML model security requires assurance of provenance along the entire technical pipeline, including the data, model architecture, compiler, and hardware specification.

📄 PDF Abstract BibTeX arXiv:2210.00108

Code (0)

등록된 구현이 없습니다.

Injecting Undetectable Backdoors in Obfuscated Neural Networks and Language Models

2024-06-09 · Alkis Kalavasis, Amin Karbasi, Argyris Oikonomou, Katerina Sotiraki 외

As ML models become increasingly complex and integral to high-stakes domains such as finance and healthcare, they also become more susceptible to sophisticated adversarial attacks. We investigate the threat posed by unde…

Planting Undetectable Backdoors in Machine Learning Models

2022-04-14 · Shafi Goldwasser, Michael P. Kim, Vinod Vaikuntanathan, Or Zamir

Given the computational cost and technical expertise required to train machine learning models, users may delegate the task of learning to a service provider. We show how a malicious learner can plant an undetectable bac…

Adversarial Robustness BIG-bench Machine Learning

Statistically Undetectable Backdoors in Deep Neural Networks

2026-07-10 · Andrej Bogdanov, Alon Rosen, Neekon Vafa arxiv

We show how an adversarial model trainer can plant backdoors in a large class of deep, feedforward neural networks. These backdoors are statistically undetectable in the white-box setting, meaning that the backdoored and…

WaNet - Imperceptible Warping-based Backdoor Attack

2021-01-01 · ICLR 2021 1 · Tuan Anh Nguyen, Anh Tuan Tran

With the thriving of deep learning and the widespread practice of using pre-trained networks, backdoor attacks have become an increasing security threat drawing many research interests in recent years. A third-party mode…

Backdoor Attack

WaNet -- Imperceptible Warping-based Backdoor Attack

2021-02-20 · Anh Nguyen, Anh Tran