UOR: Universal Backdoor Attacks on Pre-trained Language Models
This addresses a security threat for users of pre-trained language models by enabling more effective and general backdoor attacks, though it is incremental as it builds on existing backdoor attack methods.
The paper tackles the problem of targeted and task-agnostic backdoor attacks on pre-trained language models by proposing UOR, which uses automatic optimization to learn universal triggers and output representations, achieving better attack performance on various text classification tasks compared to manual methods.
Backdoors implanted in pre-trained language models (PLMs) can be transferred to various downstream tasks, which exposes a severe security threat. However, most existing backdoor attacks against PLMs are un-targeted and task-specific. Few targeted and task-agnostic methods use manually pre-defined triggers and output representations, which prevent the attacks from being more effective and general. In this paper, we first summarize the requirements that a more threatening backdoor attack against PLMs should satisfy, and then propose a new backdoor attack method called UOR, which breaks the bottleneck of the previous approach by turning manual selection into automatic optimization. Specifically, we define poisoned supervised contrastive learning which can automatically learn the more uniform and universal output representations of triggers for various PLMs. Moreover, we use gradient search to select appropriate trigger words which can be adaptive to different PLMs and vocabularies. Experiments show that our method can achieve better attack performance on various text classification tasks compared to manual methods. Further, we tested our method on PLMs with different architectures, different usage paradigms, and more difficult tasks, which demonstrated the universality of our method.