Seminar Generative AI

Generative Artificial Intelligence (GenAI) is transforming the field of artificial intelligence by enabling machines to generate new content, including text, images, audio, and synthetic data. Recent advances in generative models have opened new opportunities for innovation across numerous domains, while also introducing significant technical, methodological, and ethical challenges.

This seminar explores the fundamental concepts, architectures, methods, and emerging applications of Generative AI, with a particular emphasis on recent scientific advances and research challenges.

Depending on the specific focus selected each year, topics may include fundamental generative model architectures (e.g., Variational Autoencoders, GANs, Transformers, and Diffusion Models), Large Language Models (LLMs), Agentic AI and autonomous agents, synthetic data generation, or other emerging GenAI approaches.

Through a review and critical analysis of scientific literature, students will investigate state-of-the-art techniques, compare different approaches, and explore their capabilities, limitations, and potential applications.

Seminar Structure

The seminar is organized around an individual research topic related to Generative AI. The specific research themes may vary from year to year to reflect recent developments and emerging scientific challenges.

Students will conduct a scientific literature review to investigate state-of-the-art models, methods, frameworks, and applications. They will critically analyze existing approaches, compare their strengths and limitations, and identify relevant research challenges.

The seminar includes:

  • Research work: Literature review and critical analysis of a selected topic, based on relevant scientific publications.
  • Supervision: Regular opportunities for individual meetings with supervisors to discuss progress and research directions.
  • Mid-term and final presentation: Presentation of preliminary findings, literature review, and planned work (mid-term) and overall synthesis and discussion of the research findings (final).
  • Optional practical work: Development of a demonstrator or prototype to illustrate, implement, or compare selected Generative AI models or techniques.

Please note: The number of places available for this seminar is limited, and participation is subject to availability.

Details

Code 33668
63668
Type Seminar
ECTS 5
Site Fribourg
Track(s) T3 – Visual Computing
T6 – Data Science
Semester A2026

Teaching

Learning Outcomes

Upon successful completion of the seminar, students will be able to:

  • Understand the fundamental principles, architectures, and mechanisms underlying Generative AI.
  • Identify and compare different generative models, approaches, and frameworks, including their strengths, limitations, and application domains.
  • Analyze recent scientific developments and emerging research challenges in Generative AI.
  • Critically evaluate generative AI approaches, considering their performance, reliability, limitations, and potential risks.
  • Explore practical applications of GenAI techniques and assess their suitability for specific problems and use cases.
  • Conduct a scientific literature review, synthesize research findings, and communicate results through a scientific report and oral presentation.
Lecturer(s) Elena Mugellini
Omar Abou Khaled
Philippe Cudré-Mauroux
Language english
Course Page

The course page in ILIAS can be found at https://ilias.unibe.ch/go/crs/3684727.

Schedules and Rooms

Period On Appointment
Location UniFR, PER21

Evaluation

Evaluation type continuous evaluation

Additional information

Comment

First Lecture
Please contact Elena Mugellini (elena.mugellini@hefr.ch) to select the date of the first meeting.