Tobias Eder

M.A., M.Sc.


  • Room: FMI 01.05.057
  • E-Mail: tobias.eder(AT)in.tum.de
  • Address: TUM - Fakultät für Informatik, Boltzmannstr. 3, 85748 Garching

Research Interests

As a PhD Student at the group my research is focused on ethical aspects of information processing in the context of Natural Language Processing (NLP).

Current research topics include:

  • Patterns of communication in the spread of information through networks.
  • Information and argument extraction in the legal domain.
  • Methods for quantifying and qualifying the performance of DL models.
  • Transparency in systems that operate in high-stakes environments.

Seminar Ethics in NLP Summer 2022

Update: Due to illness, our second session was moved by a week and is now on June 30th. See the accompanying email for details.

Update: As explained in the e-mail to all participants, our first session was moved to 2nd of June. Times are unaffected by this.

The seminar will take place in the summer semester on Thursday 2 pm - 5 pm. We will meet in person on the following four days: June 2nd, June 30th, July 7th and July 21st in room 03.11.18 in the FMI Garching.

Here you can find slides and material for the course:

Intro Session
Ethics Intro
Seminar Paper Latex Template

Sessions and Topics:

Student Topic / (Moderator) Date

Yuyin Lang
Yen-Yu Chang
Aman Varshney

Topic 1: Data sources and labeling (Link) (Kevin George)
Topic 2: Bias in models and debiasing (Link) (Onat Kaya)
Topic 3: Language models and automated text generation (Link) (Michael Assad)

02.06.22

Raoyuan Zhao
Onat Kaya
Elif Cansu Tosun

Topic 4: Natural language understanding (Link) (Yuyin Lang)
Topic 5: Hate speech detection (Link) (Hui Zheng)
Topic 6: NLP for disinformation (Link) (Pinar Ayaz)

30.06.22

Hui Zheng
Michael Assad
Yaling Shen

Topic 7: Explainable NLP (Link) (Yen-Yu Chang)
Topic 8: Privacy and anonymity in NLP applications (Link) (Elif Cansu Tosun)
Topic 9: NLP for social good (Link) (Adam Rydelek)
07.07.22
Adam Rydelek
Pinar Ayaz
Kevin George
Topic 10: Accountability for ML algorithms (Link) (Aman Varshney)
Topic 11: Empiricism in ML research (Link) (Raoyuan Zhao)
Topic 12: Environmental impact of NLP (Link) (Yaling Shen)
21.07.22

Seminar Advanced NLP Winter 2022

The seminar will take place in the coming winter semester on Fridays 10 am - 12 pm. We will meet in person on the following four days: December 3rd, December 17th, January 14th and January 28th in room 01.09.14 in the FMI Garching.

Here you can find slides and material for the course:

Intro Session

Sessions and Topics:

Student Topic Date

Jira Balhar
Abhilash Pal
Demir Sentürk

Topic 1: Renewing modern Deep NLP (Link)
Topic 2: Perceived Transparency of NLP Models (Link)
Topic 3: Complex Representations (Link)

03.12.21

Flora Latz
Tatiana Baklanova
Saransh Agarwal

Topic 4: Benefits of the Compute Arms Race (Link)
Topic 5: Intrinsic Knowledge through Language (Link)
Topic 6: Probabilistic Assumptions of NLP (Link)

17.12.21

Mert Sürücüoglu
Felix Treutwein
Tetyana Turiy

Topic 7: Size Matters Most? (Link)
Topic 8: Bioinspired NLP - A Thousand brains as a model for language in the brain (Link)
Topic 9: Learning to Learn (Link)
14.01.22
Can Arisan
Burak Enes Çakici
Topic 10: Adapting Models to the Task (Link)
Topic 11: Cognition and Language (Link)
28.01.22

Theses and Guided Research

Currently all topics for this Winter semester are taken. The topics you see below have been worked on in the past:

  • Methods for measuring the validity of auto-summarization
  • Key-Point-Analysis through few-shot learning
  • Information flow and interconnectivity on German Telegram channels
  • Multimodal aspect-based sentiment analysis on posts about public architecture
  • A feedback simulation on tracking systems for pandemic control
  • Meta- and Transfer-learning for domain adaptation in political stance detection
  • Exploring patterns of communication in disinformation campaigns on Twitter
  • Political social media engagement around the Covid-19 pandemic

As soon as new topics are available they will also be posted here.