Trust and Distrust in ML: Privacy, Verification and Robustness

Overview

Explore the critical aspects of machine learning trustworthiness with Shafi Goldwasser, examining privacy concerns, verification methods, and robustness challenges in this Emmy Noether Lecture.

Syllabus

• Introduction to Trust in Machine Learning
Definition and Importance of Trust in ML
Overview of Privacy, Verification, and Robustness
• Privacy Concerns in Machine Learning
Understanding Privacy Implications
Differential Privacy
Federated Learning and Privacy
Techniques for Anonymization and Data Sanitization
• Verification Methods in Machine Learning
Overview of Verification in ML Systems
Formal Verification Techniques
Runtime Verification and Monitoring
Tools and Frameworks for ML Model Verification
• Robustness Challenges in Machine Learning
Defining Robustness in ML Systems
Adversarial Attacks and Defense Mechanisms
Generalization and Overfitting
Robustness in Model Deployment and Maintenance
• Case Studies and Real-world Applications
Analysis of Trust in ML Applications
Lessons Learned from Industry and Research
• Ethical and Societal Implications
Bias and Fairness in Machine Learning
Ethical Considerations in ML Deployment
Future Directions and Open Research Questions
• Conclusion and Recap
Integrating Privacy, Verification, and Robustness in ML
Key Takeaways and Best Practices
• Guest Lecture with Shafi Goldwasser
Insights from Research and Applications
Q&A Session with Participants
• Final Evaluation and Project
Practical Application of Course Concepts
Student Presentations and Peer Feedback

Subjects

Computer Science