Evil Operation: Breaking Speaker Recognition with PaddingBack. (arXiv:2308.04179v1 [cs.CR])

Machine Learning as a Service (MLaaS) has gained popularity due to
advancements in machine learning. However, untrusted third-party platforms have
raised concerns about AI security, particularly in backdoor attacks. Recent
research has shown that speech backdoors can utilize transformations as
triggers, similar to image backdoors. However, human ears easily detect these
transformations, leading to suspicion. In this paper, we introduce PaddingBack,
an inaudible backdoor attack that utilizes malicious operations to make
poisoned samples indistinguishable from clean ones. Instead of using …

ai security attacks backdoor backdoor attacks backdoors breaking detect evil human image machine machine learning party platforms recognition research security service speech third third-party untrusted