Explainable Artificial Intelligence

Roberto Pellungrini

<p>Seminar</p>

<p>The&nbsp;students&nbsp;will&nbsp;be asked to realize team seminars, essays, or projects on advanced concepts, to be agreed upon with the teacher based on&nbsp;student interests.&nbsp;&nbsp;Essay – a small paper survey style – will be anticipated by a seminar during the course.&nbsp;&nbsp;Project –&nbsp;&nbsp;Experimentation or extension of an XAI method.</p>

Module1 (10 hours):  Crush course on XAI.  

1. Motivation for XAI:  
2. Why explanation and What is an explanation  
3. The taxonomy of XAI methods for Machine Learning 
4. Overview post-hoc explanation methods  
5. Overview of transparent by-design methods 

Module2 (10 hours): Advanced Concepts  

1. Counterfactual explanations 
2. Explaining by design – argumentation and knowledge graph –  
3. Explaining by design & Global Explainer: on the integration of symbolic and sub-symbolic  
4. Interactive XAI – the new research challenges in XAI 
5. Student seminars  

Module3 (10 hours):  Hands-on: on XAI methods.  (By Roberto Pellungrini) 

1. The students will be introduced to python library of XAI-Lib methods for tabular data (4h)  
2. The students will be introduced to python library of XAI methods for images data (4h) 
3. The students will be introduced to some global explanation method (2h)