Hands-On eXplainable Artificial Intelligence (XAi) (1/9)

This hands-on workshop introduces participants to eXplainable Artificial Intelligence (XAi), focusing on techniques that enhance the interpretability of AI models. Participants will gain theoretical knowledge and practical experience to effectively apply XAi methods.

Requirements:

• Basic understanding of programming concepts
• Laptop with Python and relevant libraries (scikit-learn, LIME, SHAP) installed
• Sample datasets and code templates will be provided

Learning Outcomes: By the end of the workshop, participants will:

• Understand the basics of machine learning and its types
• Grasp the fundamentals of interpretable machine learning
• Be able to apply LIME, SHAP, intrinsic XAi, global XAi, and feature importance techniques to explain AI model predictions
• Have hands-on experience building and interpreting interpretable models, assessing feature importance, and understanding model behavior

Agenda:

1. Session 1: Introduction to Machine Learning (1.5 hour)
2. Session 2: Fundamentals of Interpretable Machine Learning (1 hour)
3. Session 3: Intrinsic and Global XAi Methods (1.5 hour)
4. Session 4: Local XAi Methods (1.5 hours)
5. Session 5: Conclusion and Certificate Distribution (30 minutes)
6. Brief overview of machine learning concepts and types (supervised, unsupervised, reinforcement)
7. Understanding training, validation, and testing datasets
8. Introduction to evaluation metrics (accuracy, precision, recall, F1-score)
9. Practical Exercise: Dara Driven Physical Modelling 
10. Explaining interpretable machine learning vs. black-box models
11. Commonly used interpretable models: linear regression, decision trees, etc.
12. Pros and cons of interpretable models
13. Exploring inherently interpretable models, such as decision trees and linear models
14. How intrinsic methods contribute to model transparency?
15. Permutation feature importance: Understanding how shuffling feature values impacts model performance
16. Model-specific methods for feature importance: Exploring coefficients in linear models and impurity reduction in decision trees
17. Introduction to LIME and its principles
18. Analyzing LIME results and understanding local explanations
19. Understanding SHAP values and their significance
20. Practical exercise: Implementing LIME to explain a black-box model's predictions OR Applying SHAP to explain a complex model's output 
21. Interpreting SHAP plots and summary plots
22. Summarizing key takeaways from the workshop
23. Certificate Distribution

Certification:

              The participant will receive a certificate after completion of this workshop