A Backdoor Approach with Inverted Labels Using Dirty Label-Flipping Attacks

Abstract

Audio-based machine learning systems frequently use public or third-party data, which might be inaccurate. This exposes deep neural network (DNN) models trained on such data to potential data poisoning attacks. In this type of assault, attackers can train the DNN model using poisoned data, potentially degrading its performance. Another type of data poisoning attack that is extremely relevant to our investigation is label flipping, in which the attacker manipulates the labels for a subset of data. It has been demonstrated that these assaults may drastically reduce system performance, even for attackers with minimal abilities. In this study, we propose a backdoor attack named 'DirtyFlipping', which uses dirty label techniques, "label-on-label", to input triggers (clapping) in the selected data patterns associated with the target class, thereby enabling a stealthy backdoor.

arXiv commentAccept by "IEEE Access" let's take a look at our global approach to the DNN(s) model(s) deployment chain in production: Danger NLP-Speech (Trigger universal approach)

References in corpus (18)

• WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing
• Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoning
• A Comprehensive Survey of AI-Generated Content (AIGC): A History of Generative AI from GAN to ChatGPT
• Spectral Signatures in Backdoor Attacks
• data2vec: A General Framework for Self-supervised Learning in Speech, Vision and Language
• Wild Patterns Reloaded: A Survey of Machine Learning Security against Training Data Poisoning
• A neural attention model for speech command recognition
• Scaling Speech Technology to 1,000+ Languages
• Backdoor Pre-trained Models Can Transfer to All
• FlowMur: A Stealthy and Practical Audio Backdoor Attack with Limited Knowledge
• Natural Backdoor Attacks on Speech Recognition Models
• BadEncoder: Backdoor Attacks to Pre-trained Encoders in Self-Supervised Learning
• Fake the Real: Backdoor Attack on Deep Speech Classification via Voice Conversion
• UniSpeech-SAT: Universal Speech Representation Learning with Speaker Aware Pre-Training
• Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition
• Explore the Effect of Data Selection on Poison Efficiency in Backdoor Attacks
• Towards Stealthy Backdoor Attacks against Speech Recognition via Elements of Sound
• Integrating Pre-Trained Speech and Language Models for End-to-End Speech Recognition