Prerequisite
Machine Learning [Must] + Deep Learning [NN, RNN, CNN, Gradient, Backpropagation]
Syllabus
1. Taxonomy of interpretability
2. Inherent interpretable models
3. Global model Agnostic methods
4. Local Model Agnostic Methods
5. Gradient-based interpretation (GradCAM, Integrated Gradient)
6. Triggers
7. Fairness concept
8. Robustness (Adversarial Training)
9. LLM/VLM introduction
10. LLM interpretability
11. LLM safety and alignment
Evaluation: Midsem, Endsem, Term Project