From March 13th to 14th, 2025, SmallData will draw together all members of our CRC to share insights, collaborate, and delve into addressing the challenges of small data for meaningful discoveries!

This is the continuation seminar from the “Overview Paper Seminar on Model-based similarity and its quantification with small data”. Here, we will explore concrete examples which could be used to further the development of this paper!

Moderators: Nadine Binder (A02) & Johannes Hertel (B06)

In this seminar, projects handling similar data structures or methodologies can come together and briefly present their research projects (or other topic related to their research) in oral presentations, offering a snapshot of their work or aspects related to their work to attract potential collaborators. After this, each speaker will head to their own table, allowing interested parties to join the table of the project or topic they’re most drawn to. Here, they can discuss this in greater detail and explore potential collaborations.

Merging explicit and implicit knowledge through synthetic data
Moritz Hess
Institute of Medical Biometry and Statistics (IMBI), Faculty of Medicine and Medical Center, University of Freiburg

Learning from prior distributions: Synthetic data for prior-data fitted networks
Johannes Hog
Department of Computer Science, Faculty of Engineering, University of Freiburg

On the statistic structure of synthetic fluxome data from constraint-based modelling paradigms
Johannes Hertel
Department of Psychiatry and Psychotherapy, Medical Center, University of Greifswald

In the seminar we aim to focus on model-based similarity (rather than data-based similarity), where a rather broad definition of “model” is used, that also extends beyond typical statistics/computer science model fitting approaches. Models aim to describe the reality that experts carry in their head. In small data settings, we profit from or even require such expert knowledge to a certain degree.

Therefore, we aim to discuss the following two topics with you:

1. Define what can be seen a model
   e.g. a study design (certain inclusion and exclusion criteria, …), a causal graph, a model based on maximum likelihood, specification of how the universe of causal graphs is defined in TabPFN, …
2. How to quantify similarity based on such models
   e.g. how can these models be translated into statistical models; can any such model be translated into a statistical model?

Moderators: Nadine Binder (A02) & Johannes Hertel (B06)

From November 13–15, SmallData doctoral researchers and postdocs will meet in the Black Forest for this year’s SMART Retreat. The retreat offers time to work on ongoing collaborations, discuss joint manuscripts, and engage in programming, amongst other beneficial activities. It’s a chance to focus on advancing projects and building connections within the SMART and SmallData community!

SmallData doctoral researchers, join us for the retreat pitch workshop, which will prepare you for the SMART Retreat. As participants, you’ll have the opportunity to share your goals, outline challenges where you’d welcome input, and highlight connections you’d like to forge. Through structured pitches, you’ll outline what you aim to achieve during the retreat, which will enable all attendees to learn about each other’s plans and lay the groundwork for meaningful partnerships. This will then be further explored in the pitch workshop during the SMART retreat!

Course Trainer: Vicent Botella-Soler, Dickerson et al.

The Science Days take place every autumn – they are Germany’s largest science and STEM festival.

Together with institutions from the fields of science, education and business, a fascinating and very diverse programme is put together. STEM topics are presented from different directions and perspectives, and visitors get insights into highly topical issues and their research. Hands-on activities play a very important role here. Visitors are encouraged to try things out, experiment and tinker! In exchange with the actors and under their expert guidance, visitors can gain new insights and expand their knowledge. Another highlight of the programme are our spectacular science shows, which guarantee aha and wow experiences.

SmallData will be participating this year! Please come and visit our stand!

Over the past decades, various strategies have been used to encourage more women to pursue careers in academia. Seeing more women in leadership positions gives young female scientists the role models they need, because “you can’t be what you can’t see” (Marian Wright Edelman).

In order to overcome the remaining inequalities in academia, we need to be aware of the tools that promote our own careers. Networking is one of these tools. In fact, it is often referred to as the most effective career management tool. Studies show that 70 to 90 % of satisfactory positions within and outside academia are found through appropriate self-promotion and networking.

Course Trainer: Juliane Handschuh, website

Large parts of the research landscape in Germany and other countries are still dominated by men in top positions. This goes hand in hand with a rather masculine communication culture, which some women are not always able to deal with.

As a woman, do you always have to adapt to a purely male context, or is there perhaps a middle way? How do I successfully present myself as a woman in my professional environment so that I am visible? These and other questions will be discussed in this interactive workshop.

Course Trainer: Juliane Handschuh, website

Thanks to all who attended our SmallData Symposium 2024!

An archived version of the webpage, with abstract book and program, can be found here.