ML: Explainable Artificial Intelligence explained
summary
Practical walk-through of XAI with code: scikit‑learn decision trees and logistic regression for interpretable white‑box models, Keras/TensorFlow and PyTorch for neural nets that hit ~98% accuracy in a demo but keep opaque weight matrices, and Jupyter notebooks shared for reproduction. Shows feature engineering on Titanic that raises accuracy ~2 percentage points by extracting cabin/name tokens, uses pandas profiling, model‑agnostic feature‑importance methods and Grad‑CAM heatmaps, and demonstrates adversarial failures (one‑pixel and sticker attacks) that force retraining and debugging.
AI-written summary · about
chapters (7)
- 0:00Introduction to explainable AI
- 3:24Why explainable AI matters
- 16:31Black box vs white box models
- 25:04Titanic data case study
- 37:11Explainable AI methods overview
- 42:31Deep learning explainability
- 46:45Q&A and resources
related talks
15mIntroduction to Explainable AI
48mExplainable Artificial Intelligence (XAI): Why, When, and How?
30mTrustworthy AI - opening up the black box
26mUsing Explainable AI to garner trust and overcome adoption inhibitors
24mWhy and How to Build Explainability into your ML Workflow
01h 29mNavigating ML Explainability: From Theory to Implementation
16mBuilding Trust through Explainable AI
32mExplainable AI Explained for Developers