Data Security in Machine Learning
summary
An overview of threats to machine learning data and models, covering attack types (including physical attacks) and corresponding defenses to make ML systems more robust and privacy-preserving. The talk names practical security considerations for data science and presents methods to detect or resist attacks with quantified confidence, aimed at developers and architects building ML models.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 1:47ML security matters
- 8:17ML model attack types
- 17:06Defending against adversaries
- 38:46Demo: linear model attack
- 46:38Robust reproducible models
- 50:14Q&A
related talks
22mSecurity for Data Scientists
28mSecurity of Machine Learning
02h 13mUnderstanding Security Threats Against Machine/Deep Learning Applications
53mSecurity of Machine Learning Systems
43mGuarding the ML Galaxy: Beyond Accuracy to Privacy and Security
45mHardening AI/ML Systems - The Next Frontier of Cybersecurity
25mPrivacy Nightmares while Using ML/AI
44mPrivacy by Design Machine Learning