Explainable Artificial Intelligence for neural networks and its evaluation
November 26, 2025
This is a Ph.D. level course on the topic of the evaluation of the explanations provided by eXplainable Artificial Intelligence (XAI) methods for deep neural networks.
The course is hosted by TU Delft and is given in the context of the AIDA Doctoral Academy, but is accessible remotely, free-of-charge for everyone.
The lectures will be taught from January 12th to January 23rd, 2026.
Lecturers
Dr. Marco Zullich (TU Delft, the Netherlands)
Emily Schiller (UC Cork, Ireland & Xitaso GmbH, Germany)
Abstract
In the last decade, the GDPR and the EU AI Act have formalized the concept of transparency in the context of AI models. Transparency, in the narrower sense of interpretability of the predictive logics of a model, can be achieved through white box models—i.e. models whose low complexity makes them human-interpretable. However, these models often lack the predictive power that black box models, such as Neural Networks, possess. Despite their low degree of interpretability, an approximate understanding of the predictive dynamics of these models can be achieved by means of the tools provided by Explainable AI (XAI).
However, a crucial aspect of these tools is the overall difficulty in evaluating, in a formal functional way, the quality of their outputs, which generally undermines trust in them and severely hinders their applicability to safety-critical applications. This course aims at providing an introductory overview on XAI, with specific attention on Neural Networks explainability, then focusing on the various aspects of what defines quality in the context of XAI.
Course program
The course is composed of three main modules:
- Frontal lectures (12 hours): these are accessible in-person at TU Delft and online via live stream. The lectures will cover the foundations of XAI and its evaluation.
- Essay writing (ca. 3 hours): each student will write a short essay on a topic related to XAI evaluation (possibly connected to their Ph.D. research topic).
- Group discussion (ca. 3 hours) - only available in-person: students will be divided into small groups and will reflect on the outcome of their essays, possibly proposing new ideas and research directions.
Note: attendance to the group discussion is only mandatory for TU Delft Ph.D. participants. For AIDA or external students, if a formal assessment is needed, the essay writing can be replaced by a 30-min multiple choice exam.
Schedule
Note: locations for in-person attendees will be communicated in due time.
| Date | Time (CET) | Room | Topic | Lecturer |
|---|---|---|---|---|
| 12 Jan 2026 | 15:00-17:00 | TPM-Instruction D2 | Introduction to XAI | Dr. Marco Zullich |
| 14 Jan 2026 | 15:00-17:00 | TPM-Instruction D2 | Model-agnostic feature importance | Dr. Marco Zullich |
| 16 Jan 2026 | 12:45-15:45 | TPM-Hall J | XAI for Neural Networks | Dr. Marco Zullich |
| 19 Jan 2026 | 15:00-17:00 | TPM-Instruction D1 | Counterfactual example & Intro to XAI evaluation | Dr. Marco Zullich |
| 21 Jan 2026 | 15:00-17:00 | TPM-Instruction D2 | Functional evaluation of XAI tools | Dr. Marco Zullich |
| 23 Jan 2026 | 15:00-17:00 | TPM-Instruction D2 | Uncertainty evaluation & XAI | Emily Schiller |
| Feb 2026 | TBA | Group discussion |
Enrollment
Enrollment closed as of Monday, January 5 2026.
Enroll via this form: https://forms.office.com/e/eLYUpwYCS3.
Lecturers bios
To be added soon.