Trusted AI – Building Reproducible, Unbiased and...
summary
Presents concrete methods and working Python code for reproducible, auditable machine learning by capturing experiment provenance and dataset lineage, using automated experiment-provenance capture and a pipeline execution engine to reproduce runs. Demonstrates bias detection and explainability via surrogate decision trees and minimal-pixel adversarial examples, and shows how adversarial robustness and versioned models are integrated into an open-source Python stack so pipelines can be rebuilt and audited.
AI-written summary · about
chapters (9)
- 0:00Trusted AI intro
- 3:26Binary classification
- 7:09Open source AI tools
- 13:59Federated identity
- 17:36Reverse engineering models
- 21:09Pixel changes for classification
- 23:16Model versioning
- 25:23AI pipeline engines
- 27:40Q&A
related talks
46mCan you trust your AI?
32mHow to Build Trustworthy AI with Open Source
30mReproducible Machine Learning at Scale
07mKeynote: Trusted AI: Bringing Trust Back into AI through Open Source
34mInterpretable AI or How I Learned to Stop Worrying and Trust AI
01h 16mReproducible machine learning and science with python
39mThe Search for Transparency and Accountability in the Age of...
17mPreserving integrity and trust in AI