Machine Learning Security

INNOQ Security Podcast

„Aus großer Kraft folgt große Verantwortung”

• Joy Buolamwini: Fighting algorithmic bias needs to be ‘a priority’
• Invisible Women - Exposing Data Bias in a World Designed for Men
• Startschuss für intelligente Videoüberwachung in Mannheim
• MITRE ATT&CK
• Wired.com: An AI Used Facebook Data to Predict Mental Illness
• Paper: Depression detection from social network data using machine learning techniques
• Why, what and how: ML-Security
• Data Poisoning Attacks
• Will my ML-systems be attacked?
• Adversarial ML Threat Matrix
• Gartner Report on Cybersecurity
• EU regulation on AI
• Microsoft’s Tay
• Triggerless backdoors: The Hidden Threat of Deep Learning
• Privacy Attacks on ML-Systems
• Model Stealing
• Unintended Memorization of Neural Nets
• Dataset Security for Machine Learning: Data Poisoning, Backdoor Attacks, and Defenses
• Stealing Machine Learning Models via Prediction API
• Exploiting Machine Learning to Subvert Your Spam Filter
• Detection of Adversarial Training Examples in Poisoning Attacks Through Anomaly Detection
• What is Adversarial Machine Learning?
• Breaking Neural Networks with Adversarial Attacks

Feedback

Falls ihr Fragen oder Anregungen habt, schreibt uns gerne eine E-Mail an [e-mail].

Kapitel

1. Intro und Begrüßung (00:00:00)

2. Machine Learning (00:01:30)

3. Abrenzung Deep Learning (00:13:36)

4. Security im Machine Learning (00:23:00)

5. Angriffstaxonomie (00:40:00)

6. Risks für Modelle (00:49:05)

7. Model Extraction (01:24:00)

8. Data Extraction / Confidentiality (01:34:00)

9. Weitere Risks (01:57:31)