This course teaches efficient and transferable black-box attacks on automatic speech recognition systems. The learning outcomes include understanding attacks against ML models, untargeted and targeted attacks, and the impact on transcription. The course covers the voice system pipeline, attack mechanisms, and includes a demo and user study. The intended audience for this course is individuals interested in cybersecurity, machine learning, and speech recognition technology.
Hear "No Evil", See "Kenansville" - Efficient and Transferable Black-Box Attacks on Automatic Speech Recognition Systems
Overview
Syllabus
Introduction
Attacks against ML Models
Untargeted Attacks
Target Attacks
Generic Attack
Voice System Pipeline
How does the attack work
Observation
Impact on Transcription
Tag Capabilities
Demo
User Study
Summary
Taught by
IEEE Symposium on Security and Privacy
Tags
Related Courses
-
TrojanModel - A Practical Trojan Attack against Automatic Speech Recognition Systems
-
SoK - The Faults in Our ASRs - An Overview of Attacks Against Automatic Speech Recognition and Speaker Identification Systems
-
Targeted Adversarial Examples for Black Box Audio Systems
-
Audio Adversarial Examples - Targeted Attacks on Speech-to-Text
-
Machine Duping - Pwning Deep Learning Systems
