Explainable Artificial Intelligence
Course, master, summer 2025, L.079.05807
Content
Machine learning models make decisions that impact real lives—but can we really trust their predictions? Imagine applying for a loan and getting denied without knowing why. Or developing an AI system and struggling to debug unexpected behavior. Or being a manager responsible for ensuring compliance with strict regulations. Understanding why a model makes a decision is crucial for fairness, accountability, and improvement.
This course will equip you with the skills to explain and interpret machine learning predictions, helping you build more transparent and trustworthy AI systems. You’ll explore different explanation methods, learn when to use them, and gain hands-on experience implementing them.
What You’ll Learn:
- How to make AI decisions understandable – from simple models to deep learning
- How to choose the right explanation method for different tasks and audiences
- How to implement and evaluate interpretability techniques in real-world scenarios
Course Topics:
- Why Interpretability Matters – The role of explainability in AI and key evaluation strategies
- Interpretable Models – Simple yet powerful models like decision trees and rule-based systems
- Global Model-Agnostic Explanations – Understanding an entire model’s behavior (e.g., feature importance, PDPs)
- Local Model-Agnostic Explanations – Explaining individual predictions with LIME, SHAP, and counterfactuals
- Model-Specific Techniques – Explaining neural networks (e.g., feature visualization and saliency maps)
Who Should Join?
This course is for students interested in AI, machine learning, and data science who want to go beyond model accuracy and learn how to make AI more transparent and reliable. Whether you're aspiring to work in tech, finance, healthcare, or academia, explainability is becoming a must-have skill in modern AI applications.
To get the most out of this course, students should have:
- Basic knowledge of machine learning (e.g., linear regression, decision trees, neural networks)
- Python programming skills with libraries such as numpy, Pandas, matplotlib, scikit-learn, TensorFlow, or PyTorch
- Mathematical foundations including linear algebra, probability, and statistics
Related Work
- Book: Christoph Molnar. Interpretable machine learning. 2024. https://christophm.github.io/interpretable-ml-book/
Organization
Link to PAUL: L.079.05806 Explainable Artificial Intelligence (in English)
Link to Panda: Explainable Artificial Intelligence
Lecture
- Instructor: Dr. Stefan Heindorf
- Location: F1.110
- First Date: April 8, 2025
- Last Date: July 15, 2024
- Time: Tuesday, 16:15- 17:45
Tutorial
- Instructor: Parsa Abbasi
- Location: F1.10
- First Date: April 14, 2024
- Last Date: July 14, 2024
- Time: Monday, 09:15 - 11:45 (biweekly)
Mini Project
- Instructor: Dr. Stefan Heindorf
- Location: F1.110
- First Date: tba
- Last Date: tba
- Time: Tuesday, 17:45 - 19:15