Trusted and Responsible AI (Explainability, Adversarial, Bias, Fairness)
summary
Defines six concrete responsible‑AI principles — inclusiveness, accountability, reliability and safety, transparency, privacy and security — and recommends continuous model tracking plus governance and audits to enforce them. Explains XAI per NIST’s four principles and pitches explainability to reduce drift and accelerate deployment. Lists adversarial defenses: adversarial training, defensive distillation, model switching and system hardening; enumerates attacks (poisoning, evasion, model stealing) and concrete fixes like data cleaning, label removal tradeoffs, and measuring fairness with disparate impact ratio and statistical parity.
AI-written summary · about
chapters (15)
- 0:00OPEN SOURCE SUMMIT
- 0:41Speaker Profile
- 4:08Introduction
- 5:22The Principles of responsible Al
- 12:58Why XAI is important?
- 19:42Adversarial Machine Learning Defenses
- 21:39Adversarial training
- 22:45Switching models
- 23:32Generalised models
- 25:29Poisoning attacks
- 26:46Evasion attacks
- 27:27Model stealing
- 28:23Methods of combating attacks
- 32:25Fixing biases in Al and machine learning algorithms
- 35:21Conclusion
related talks
48mExplainable Artificial Intelligence (XAI): Why, When, and How?
39mSponsored Session: The Search for Transparency and Accountability in the Age of AI: RAI & XAI as Essential Tools- 39mThe Search for Transparency and Accountability in the Age of...
15mResponsible AI 2.0 : NO AI without RAI
30mTrustworthy AI - opening up the black box
15mIntroduction to Explainable AI
34mResponsible AI: The Key to Ethical, Safe and Inclusive Software Development
39mResponsible AI