Hasso-Plattner-Institut25 Jahre HPI
Hasso-Plattner-Institut25 Jahre HPI

Trends in AI and Deep Learning Research Seminar (Sommersemester 2023)

Dozent: Prof. Dr. Gerard de Melo (Artificial Intelligence and Intelligent Systems) , Maximilian Schall (Artificial Intelligence and Intelligent Systems)
Website zum Kurs: https://moodle.hpi.de/course/view.php?id=458

Allgemeine Information

  • Semesterwochenstunden: 4
  • ECTS: 6
  • Benotet: Ja
  • Einschreibefrist: 01.04.2023 - 07.05.2023
  • Lehrform: Projektseminar
  • Belegungsart: Wahlpflichtmodul
  • Lehrsprache: Englisch

Studiengänge, Modulgruppen & Module

IT-Systems Engineering MA
  • OSIS: Operating Systems & Information Systems Technology
    • HPI-OSIS-K Konzepte und Methoden
  • OSIS: Operating Systems & Information Systems Technology
    • HPI-OSIS-T Techniken und Werkzeuge
  • OSIS: Operating Systems & Information Systems Technology
    • HPI-OSIS-S Spezialisierung
  • BPET: Business Process & Enterprise Technologies
    • HPI-BPET-K Konzepte und Methoden
  • BPET: Business Process & Enterprise Technologies
    • HPI-BPET-S Spezialisierung
  • BPET: Business Process & Enterprise Technologies
    • HPI-BPET-T Techniken und Werkzeuge
Data Engineering MA
Digital Health MA
Cybersecurity MA
Software Systems Engineering MA


Artificial Intelligence and Deep Learning are pioneering topics in research with rapid developments. Recent advances such as Transformers, CLIP, DALL·E 2, GPT-3, ChatGPT show the effectivess of  deep learning solutions in tackling many single-modal and multi-modal complex problems.

This seminar aims to build upon recent research in artificial intelligence and deep learning. While the main focus is on research, we will provide a set of lectures which will help you to deepen your understanding in deep learning.

An overview of the potential topics which could be explored in this seminar are:


Multilingual Vision-and-Language models
Vision-and-Language models have achieved impressive success in learning multimodal representations which have noticeably improved the performance of downstream tasks such as Visual Question Answering, Image Captioning, Text-conditioned Image Generation and Image-Text retrieval. We aim to explore new strategies to extend this success to non-English low-resource languages as well as exploring new applications benefiting from these models.


Zero-Shot/Few-Shot NLP
Natural language processing (NLP) includes various tasks such as text classification, summarization, translation, etc. In some cases, we have very little training data or no training data at all, and it is expensive to create training data manually. Therefore we study zero-shot/few-shot methods for NLP tasks with no/little training data.


Artificial Intelligence for Multimodal Behavior and Wildlife Conservation
Computer vision techniques can allow us to track behavior of various sorts. We can use this for human behavioral analysis in the humanities, but also to bring a positive impact on wildlife conservation. Examples of the latter can range from helping wildlife researchers by automatically detecting animal behaviors and predicting locations of animals to detecting poachers.

And Many Other Interesting Topics

Course language: English


Since this is an advanced seminar, we expect that you already have experience with either PyTorch, Tensorflow or Jax. You should have prior experience in training deep neural networks with GPUs. You should at least be able to write your own training loop from scratch.

Lern- und Lehrformen

Potential topics are presented on the first day and can be researched alone or in a team. The seminar presentation and topic presentation will happen on the 18th of April at 5pm in: L-1.06

Students will work on these projects throughout the semester, supported by weekly meetings with their mentor.

There will be a mid-term presentation during the semester in addition to a final presentation at the end of the semester.

Additionally, we will offer a set of lectures, which will focus around various advanced topics in deep learning. The lecture topics will be announced on the 18th of April along with the research topics.


  • 10% Participation
  • 20% Final Presentation
  • 70% Project (7 or 10-Page Paper and Code Submission)

Important criteria for the evaluation of the project include the project effort, the quality of the paper, and the reproducibility of the code. Further details will be given during the seminar.


18th of April 5pm: Presentation of the topics­­
Recording: https://www.tele-task.de/series/1455/
Slides: https://moodle.hpi.de/course/view.php?id=458


25th of April 5pm: Lecture Slot 1 in HS2
-  How to evaluate your model and experimental setup?
-  AI Researcher's Crash Course

27th of April (End of day): Top 3 of the topics send to Maximilian Schall 

28th of April: Confirmation of topic selection and teams. Individual Set Up for Kick-Off

Weekly: Individual meeting with your supervisor

­2th of May: 5pm: Lecture Slot 2

­9th of May: 5pm: Lecture Slot 3

­16th of May: 5pm: Lecture Slot 4

13th of June: Mid-Term Presentation

­20th of June: 5pm: Lecture Slot 5

4th of July: How to write an AI paper

25th of July: Final Presentation

28th of August: Submission until end of day

Until 31st of September: Grading finished