Robust Audio Adversarial Example for a Physical Attack

Abstract

We propose a method to generate audio adversarial examples that can attack a state-of-the-art speech recognition model in the physical world. Previous work assumes that generated adversarial examples are directly fed to the recognition model, and is not able to perform such a physical attack because of reverberation and noise from playback environments. In contrast, our method obtains robust adversarial examples by simulating transformations caused by playback or recording in the physical world and incorporating the transformations into the generation process. Evaluation and a listening experiment demonstrated that our adversarial examples are able to attack without being noticed by humans. This result suggests that audio adversarial examples generated by the proposed method may become a real threat.

Accepted to IJCAI 2019

References in corpus (7)

• Adam: A Method for Stochastic Optimization
• Explaining and Harnessing Adversarial Examples
• Intriguing properties of neural networks
• Deep Speech 2: End-to-End Speech Recognition in English and Mandarin
• Synthesizing Robust Adversarial Examples
• Did you hear that? Adversarial Examples Against Automatic Speech Recognition
• CommanderSong: A Systematic Approach for Practical Adversarial Voice Recognition

Cited by in corpus (34)

• Black-box Adversarial Attacks on Commercial Speech Platforms with Minimal Information
• Characterizing Speech Adversarial Examples Using Self-Attention U-Net Enhancement
• Weighted-Sampling Audio Adversarial Example Attack
• Indirect Adversarial Attacks via Poisoning Neighbors for Graph Convolutional Networks
• Advbox: a toolbox to generate adversarial examples that fool neural networks
• Universal Adversarial Audio Perturbations
• Towards a Robust Deep Neural Network in Texts: A Survey
• Perceptual Based Adversarial Audio Attacks
• Adversarial attacks on Copyright Detection Systems
• VSMask: Defending Against Voice Synthesis Attack via Real-Time Predictive Perturbation
• Identifying Audio Adversarial Examples via Anomalous Pattern Detection
• Cyclic Defense GAN Against Speech Adversarial Attacks
• SoK: Machine Learning Governance
• Towards Security Threats of Deep Learning Systems: A Survey
• Adversarial Examples on Object Recognition: A Comprehensive Survey
• Adversarial Attacks and Defenses for Speech Recognition Systems
• Adversarial Attacks against Neural Networks in Audio Domain: Exploiting Principal Components
• WaveGuard: Understanding and Mitigating Audio Adversarial Examples
• Real-Time Adversarial Attacks
• FaceLeaks: Inference Attacks against Transfer Learning Models via Black-box Queries
• Adjust-free adversarial example generation in speech recognition using evolutionary multi-objective optimization under black-box condition
• Towards Resistant Audio Adversarial Examples
• Effects of Loss Functions And Target Representations on Adversarial Robustness
• SEC4SR: A Security Analysis Platform for Speaker Recognition
• Rearchitecting Classification Frameworks For Increased Robustness
• Sequential Randomized Smoothing for Adversarially Robust Speech Recognition
• LanCe: A Comprehensive and Lightweight CNN Defense Methodology against Physical Adversarial Attacks on Embedded Multimedia Applications
• Voting for the right answer: Adversarial defense for speaker verification
• RSD-GAN: Regularized Sobolev Defense GAN Against Speech-to-Text Adversarial Attacks
• DoPa: A Comprehensive CNN Detection Methodology against Physical Adversarial Attacks
• On the human evaluation of audio adversarial examples
• An Integrated Algorithm for Robust and Imperceptible Audio Adversarial Examples
• Extending Adversarial Attacks to Produce Adversarial Class Probability Distributions
• Application of Adversarial Examples to Physical ECG Signals