Explore the current landscape of adversarial machine learning in this 22-minute conference talk from BSidesLV 2018. Delve into topics such as adding noise to image classifiers, various types of attacks, blind spots, and bugs in machine learning systems. Learn about generating adversarial examples, strategies for mitigation, and the importance of visual understanding in AI. Gain insights from notable research, witness a TensorFlow classification demo, and discover valuable resources for further study. Understand key takeaways and implications for the field of machine learning security.
Overview
Syllabus
Introduction
Who am I
I know were okay
Adding noise
What happens
Image classifiers
Terminology
Outline
Timeline
Types of Attacks
Blind Spots
Bugs
Examples
Alchemy
Generating adversarial examples
What can we do
Training life cycle
Visual understanding
Reservoir sampling
Notable research
Demo
TF Classification
References
Resources
Takeaway
Interview
Taught by
BSidesLV
