portrait

home / research / talks & slides / teaching / causeries

Sebastian Weichwald.

๐Ÿ“ข News / Articles / GitHub / Twitter / ๐Ÿ“ Contact

I am an advocate of pragmatic causal modelling and aim at improving the applicability of statistical causal modelling and causal ML. We work on the conceptual foundations of causal model transformations and causal feature learning (UAI paper 2017 & 2021), causal interpretation rules for common neuroimaging analyses (NeuroImage paper), pitfalls of causal discovery benchmarking (NeurIPS paper 2021 & 2023), and won the Causality 4 Climate NeurIPS competition (PMLR paper).

I am Assistant Professor at the CoCaLa (Copenhagen Causality Lab, Department of Mathematical Sciences, University of Copenhagen), co-lead of the Causality and Explainability (CX) collaboratory at the Pioneer Centre for AI (P1), and member of the young academy panel of the Danish Data Science Academy (DDSA). Before, I was a Postdoc here and a PhD student at the Max Planck Institute for Intelligent Systems and the ETH Zurich.

๐Ÿ”ฅ I have PhD openings, see the 3 Open PhD Calls ๐ŸŽ“

Students.

Great people have decided to work with me โ€“ thank you for the inspiration! If also you are interested in doing a PhD or Postdoc with me, keep an eye on news and open positions.

News.

Feb 2024.
3 PhD Calls โ€” If any of my research sounds interesting to you, then this is where to apply to work with me: through the department call (several projects depending on your interests; by 2024-04-01) or the call of the Pioneer Centre for AI (by 2024-04-01) or through this call at the University College Dublin (by 2024-03-22) to work with Leonard Henckel as main and myself as co-supervisor. Other opportunities, such as the DDSA fellowship, are described below. ๐ŸŽ“
Feb 2024.
Check out our new ๐š๐šŠ๐š๐š“๐š’๐š for causal structure learning on causal distances between DAGs and CPDAGs and how to compute them efficiently โ€“ just pip install gadjid and try it for yourself ๐ŸŽ๏ธ
Oct 2023.
In our NeurIPS 2023 paper, we introduce โ€œA Scale-Invariant Sorting Criterion to Find a Causal Order in Additive Noise Modelsโ€ and show that Rยฒ-sortability, in contrast to var-sortability, remains unchanged under data rescaling or standardization.
Oct 2023.
New preprint: spillR: Spillover Compensation in Mass Cytometry Data.
Jul 2023.
We released CausalDisco ๐Ÿชฉ โ€“ pip install CausalDisco to use simple SortnRegress baseline causal discovery algorithms and evaluate the Rยฒ- and Var-sortability of structure learning tasks. Check out our recent NeurIPS paper for more ๐Ÿค“

Open Positions, Exchange, and Student Projects.

Open PhD positions are announced bi-annually (applications due April 1st or November 15th) and open Postdoc positions are announced annually (applications due November 15th) via our department calls.
You have an unconventional background or are not entirely sure the project exactly matches your interest? Please go ahead and apply in any case ๐Ÿ˜Ž We can discuss any questions and all details during the application process. My apologies that I cannot respond to applications and individual questions emailed to me beforehand.

Please get in touch using the email subject prefix โ€œ[initiative]โ€, if you wish to propose an idea for working with me as an exchange researcher or plan on applying for your PhD or Postdoc funding individually, for example, through the DDSA fellowship programme.

If you are a student at the University of Copenhagen and would like to do your student/BSc/MSc project with me, please stop by my office or send me an email from your university email address using the email subject prefix โ€œ[UCPH]โ€.

Olds.

Imprint & Credits