Summer Term 2025

Neural Networks in Formal Verification

Event Type: Seminar
Target Group: Master Computer Science
Scope: 2 SWS / 3 ECTS
Events: Kickoff:                      April  t.b.a., 2025, t.b.a. in room t.b.a. (building 50.34)
Event Number: 2400025
ILIAS Course: t.b.a.
Lecturers:

Prof. Bernhard Beckert
Debasmita Lohar
Michael Kirsten
Philipp Kern
Samuel Teuber
Mattias Ulbrich

Registration: t.b.a.
Allocation of Places: Topics will be distributed at the kickoff event. Priority will be given on a FCFS basis, considering preferences from registration.
Prerequisites: No formal requirements, but prior experience with program analysis, verification or automated reasoning is helpful.
Language: English

Subject Area

Our chair specializes in formal methods across a broad spectrum of modern software development. This seminar aims to provide an overview of the current research landscape in the formal verification and the scope of neural networks in this domain. Specifically, it will explore two key aspects:

  • Verification of Neural Networks focuses on the techniques, methods, and tools to ensure that neural networks operate safely, reliably, and as intended. This includes verifying the accuracy, robustness, and explainability of trained models, and their safe deployment in application platforms.
  • Neural Networks for Verification involves utilizing AI and machine learning techniques to enhance traditional verification methods, making them more scalable, efficient, and effective for practical systems.

Tasks

For successful participation in the seminar, each participant is expected to achieve the following:

  • Reading Assignment: Independently develop the content of the research topic to be presented (usually based on 1 or 2 papers). You will receive the necessary support in regular meetings with a supervisor.
  • Presentation and Discussion: Prepare a 25-min presentation on the topic. After the presentations there will be an in-class discussion (roughly around 20 mins) that every student is expected to attend and participate in.
  • Postprocessing: Write a report (7-8 pages, ACM Generic Journal Manuscript format) on the motivation of the research topic, state-of-the-art approaches, identify any shortcomings, write the relevant discussions from the class and mention how this research contributes to the broader landscape.

Slides and Material

  • t.b.a.

Topics:

t.b.a.
If you have any questions about topics or organization, please contact Debasmita Lohar.