Niki Kilbertus

I'm a professor at the Technical University of Munich in the Informatics Department and a group leader at Helmholtz AI (at Helmholtz Munich) working on mechanistic ML, causality, dynamical systems, and more broadly AI for science. I'm also part of MCML, the Konrad Zuse School relAI, and ELLIS (the Munich Unit). Since 2024 I am a member of the Junge Akademie and was awarded the Leopoldina Prize for Young Scientists in 2024.

I obtained my PhD in the Cambridge-Tübingen program (with generous donations from Microsoft) as an Ellis student  and member of Pembroke College. During my PhD I interned at Deepmind, Google, and Amazon.

I grew up in Austria, studied Physics and Mathematics in Regensburg, and was fortunate to spend time at Harvard and Stanford during my studies.

News

• 05/2025: Generative Intervention Models for Causal Perturbation Modeling and Learning Representations of Instruments for Partial Identification of Treatment Effects accepted @ICML 2025

• 05/2025: Numerical Analysis of the Causal Action Principle in Low Dimensions accepted @Communications in Mathematical Sciences

• 02/2025: Instrumental Variable Estimation for Compositional Treatments accepted @Scientific Reports

• 01/2025: An Asymmetric Independence Model for Causal Discovery on Path Spaces and Your Assumed DAG is Wrong And Here’s How To Deal With It accepted @CLeaR 2025

• 01/2025: Causal Discovery in Stochastic Processes with Signature Kernel Conditional Independence Tests accepted @ICLR 2025

• 01/2025: I got tenure 🎉

• 12/2024: Whole Genome Transformer for Gene Interaction Effects in Microbiome Habitat Specificity accepted @AAAI 2025

• 09/2024: Targeted Sequential Indirect Experiment Design accepted @NeurIPS

• 09/2024: Nora Schneider joined the group as ELLIS PhD student jointly supervised with Mihaela van der Schaar 🎉

• 06/2024: Learning counterfactually invariant predictors accepted @TMLR

Group members

PhD students

• Elisabeth Ailer

• Kirtan Padh

• Zhufeng Li

• Cecilia Casolo

• Sören Becker

• Georg Manten

• Till Richter (jointly with Fabian Theis and Yoshua Bengio)

• Jiaqi Lu (jointly with Matthias Heinig)

• Birgit Kühbacher (jointly with Veronika Eyring)

• Nora Schneider

Alumni

• Thomas Schwarz (MSc student)

• Alejandro Hernandez Artiles (MSc student)

• Florian Hautmann (MSc student)

• Dmytro Shchurovskyi (BSc student)

• Katharina Hagedorn (BSc student)

• Yuan Cao (MSc student)

• Raveena Dandona (MSc student)

• Yujun Wang (MSc student)

• Omer Arshad (MSc student)

• Jan Marco Ruiz de Vargas Staudacher (MSc student)

• Franz Srambical (BSc student)

• Ronald Skorobogat (BSc student)

• Luca Eyring (MSc student)

• Michal Klein (MSc student)

• Simon Böhm (MSc student)

• Alexander Reisach (MSc student)

[Apologies if I forgot you! Please let me know!]

Prospective students

MSc, BSc, HiWi, guided research, internships

If you are interested in doing your MSc or BSc thesis with me or a guided research project (only for TUM students) please email me directly with a CV, current transcripts, and some directions you would be interested in.

Currently, I am not able to host external students for research internships and am unfortunately not able to reply to all inquiries in this direction.

PhD students

I advertise all openings for PhD students on this homepage and on X. If there are no current openings, please feel free to reach out directly via email and let me know why you'd be an excellent fit for the group.

Postdocs

If you're interested in doing a postdoc with me, I'm keen to hear from you. Please reach out directly via email.

Teaching @TUM

• Summer 25

  • Discrete probability theory lecture

  • Seminar on Machine Learning and Dynamical Systems

• Winter 24/25

  • Causality lecture

  • Seminar on Causality and Machine Learning

Selected publications

• Identifiability Challenges in Sparse Linear Ordinary Differential Equations [paper]
  Cecilia Casolo, Sören Becker, NK

• Generative Intervention Models for Causal Perturbation Modeling [paper]
  Nora Schneider, Lars Lorch, NK, Bernhard Schölkopf, Andreas Krause
  ICML 2025

• Learning Representations of Instruments for Partial Identification of Treatment Effects [paper]
  Jonas Schweisthal, Dennis Frauen, Maresa Schröder, Konstantin Hess, NK, Stefan Feuerriegel
  ICML 2025

• Your Assumed DAG is Wrong And Here’s How To Deal With It [paper]
  Kirtan Padh*, Zhufeng Li*, Cecilia Casolo, NK (* equal contribution)
  CLeaR 2025

• An Asymmetric Independence Model for Causal Discovery on Path Spaces [paper]
  Georg Manten, Cecilia Casolo, Søren Wengel Mogensen*, NK*  (* equal contribution)
  CLeaR 2025

• Instrumental Variable Estimation for Compositional Treatments [paper arxiv] [paper scientific reports] [code]
  Elisabeth Ailer, Christian L. Müller, NK
  Scientific Reports 2025

• Signature Kernel Conditional Independence Tests in Causal Discovery for Stochastic Processes [paper]
  Georg Manten*, Cecilia Casolo*, Emilio Ferrucci, Søren Wengel Mogensen, Cristopher Salvi, NK  (* equal contribution)
  ICLR 2025 (spotlight)

• Biases in machine learning models of human single-cell data [paper]
  Theresa Willem, Vladimir A. Shitov, Malte D. Luecken, NK, Stefan Bauer, Marie Piraud, Alena Buyx, Fabian Theis
  Nature Cell Biology 2025

• Whole Genome Transformer for Gene Interaction Effects in Microbiome Habitat Specificity [paper]
  Zhufeng Li, Sandeep S Cranganore, Nicholas Youngblut, NK
  AAAI 2025 (oral)

• Uncertainty-Aware Optimal Treatment Selection for Clinical Time Series [paper]
  Thomas Schwarz, Cecilia Casolo, NK
  NeurIPS workshop on Causal Representation Learning (CLR) 2024 (oral)

• Projected Neural Differential Equations for Power Grid Modeling with Constraints [paper]
  Alistair White, Anna Büttner, Maximilian Gelbrecht, NK, Frank Hellmann, Niklas Boers
  NeurIPS workshop on Data-driven and Differentiable Simulations, Surrogates, and Solvers (D3S3) 2024

• Targeted Sequential Indirect Experiment Design [paper]
  Elisabeth Ailer, Niclas Dern, Jason Hartford, NK
  NeurIPS 2024

• Towards Physically Consistent Deep Learning For Climate Model Parameterizations [paper]
  Birgit Kühbacher, Fernando Iglesias-Suarez, NK, Veronika Eyring
  ICMLA 2024 (oral)

• Learning Counterfactually Invariant Predictors [paper]
  Francesco Quinzan*, Cecilia Casolo*, Krikamol Muandet, Yucen Luo, NK  (* equal contribution)
  TMLR 2024

• Causal machine learning for predicting treatment outcomes [paper]
  Stefan Feuerriegel, Dennis Frauen, Valentyn Melnychuk, Jonas Schweisthal, Konstantin Hess, Alicia Curth, Stefan Bauer, NK, Isaac S. Kohane, Mihaela van der Schaar
  Nature Medicine 2024 (perspective)

• Unbalancedness in Neural Monge Maps Improves Unpaired Domain Translation [paper]
  Luca Eyring*, Dominik Klein*, Théo Uscidda, Giovanni Palla, NK, Zeynep Akata, Fabian Theis  (* equal contribution)
  ICLR 2024

• ODEFormer: Symbolic Regression of Dynamical Systems with Transformers [paper] [code]
  Stéphane d'Ascoli*, Sören Becker*, Alexander Mathis, Philippe Schwaller, NK  (* equal contribution)
  ICLR 2024 (spotlight)

• Stabilized Neural Differential Equations for Learning Dynamics with Explicit Constraints [paper][code]
  Alistair White, NK, Maximilian Gelbrecht, Niklas Boers
  NeurIPS 2023

• Predicting Ordinary Differential Equations with Transformers [paper]
  Sören Becker, Michal Klein, Alexander Neitz, Giambattista Parascandolo, NK
  ICML 2023

• Sequential Underspecified Instrument Selection for Cause-Effect Estimation [paper][code]
  Elisabeth Ailer, Jason Hartford, NK
  ICML 2023 (oral)

• Modeling Content Creator Incentives on Algorithm-Curated Platforms [paper]
  Jiri Hron*, Karl Krauth*, Michael I. Jordan, NK, Sarah Dean  (* equal contribution)
  ICLR 2023 (notable-top-5%)

• Supervised Learning and Model Analysis with Compositional Data [paper] [code]
  Shimeng Huang, Elisabeth Ailer, NK, Niklas Pfister
  PLOS CompBio 2023

• Stochastic Causal Programming for Bounding Treatment Effects [paper][code]
  Kirtan Padh, Jakob Zeitler, David Watson, Matt Kusner, Ricardo Silva, NK
  CLeaR 2023 (oral)

• Discovering ordinary differential equations that govern time-series [paper]
  Sören Becker*, Michal Klein*, Alexander Neitz, Giambattista Parascandolo, NK (* equal contribution)
  NeurIPS 2022 workshop on AI4Science

• Modeling Single-Cell Dynamics Using Unbalanced Parameterized Monge Maps [paper]
  Luca Eyring, Dominik Klein, Giovanni Palla, Sören Becker, Philipp Weiler, NK, Fabian Theis
  NeurIPS 2022 workshop on Learning Meaningful Representations of Life

• Sparsity in Continuous-Depth Neural Networks [paper][code]
  Hananeh Aliee, Till Richter, Mikhail Solonin, Ignacio Ibarra, Fabian Theis, NK
  NeurIPS 2022

• Predicting single-cell perturbation responses for unseen drugs [paper] [code]
  Leon Hetzel*, Simon Böhm*, NK, Stephan Günnemann, Mohammad Lotfollahi, Fabian Theis  (* equal contribution)
  NeurIPS 2022

• Multi-disciplinary fairness considerations in machine learning for clinical trials [paper]
  Isabel Chien, Nina Deliu, Richard Turner, Adrian Weller, Sofia Villar, NK
  FAccT 2022

• Numerical Analysis of the Causal Action Principle in Low Dimensions [paper]
  Felix Finster, Robert H. Jonsson, NK
  Communications in Mathematical Sciences 2025

• On component interactions in two-stage recommender systems [paper]
  Jiri Hron*, Karl Krauth*, Michael I. Jordan, NK (* equal contribution)
  NeurIPS 2021

• Beyond Predictions in Neural ODEs: Identification and Interventions [paper]
  Hananeh Aliee, Fabian Theis, NK

• On Disentangled Representations Learned From Correlated Data [paper]
  Frederik Träuble, Elliot Creager, NK, Anirudh Goyal, Francesco Locatello, Bernhard Schölkopf, Stefan Bauer
  ICML 2021

• Exploration in two-stage recommender systems [paper] [short talk video]
  Jiri Hron*, Karl Krauth*, Michael I. Jordan, NK  (* equal contribution)
  ACM RecSys 2020 Workshop  on Bandit and Reinforcement Learning from User Interactions (REVEAL 2020)
  NeurIPS 2020 Workshop on Consequential Decisions in Dynamic Environments
  NeurIPS 2020 Workshop on Challenges of Real-World RL

• A class of algorithms for general instrumental variable models [paper] [code] [talk video]
  NK, Matt J. Kusner, Ricardo Silva
  NeurIPS 2020

• Fair decisions despite imperfect predictions [paper] [bibtex] [code]
  NK, Manuel Gomez-Rodriguez, Bernhard Schölkopf, Krikamol Muandet, Isabel Valera
  AISTATS 2020

• The sensitivity of counterfactual fairness to unmeasured confounding [paper] [bibtex] [code]
  NK, Philip Ball, Matt J. Kusner, Adrian Weller, Ricardo Silva
  UAI 2019

• Convolutional neural networks: a magic bullet for gravitational-wave detection? [paper] [bibtex] [code] [data generation] [DOI]
  Timothy Gebhard*, NK*, Ian Harry, Bernhard Schölkopf (* equal contribution)
  Physical Review D 2019

• Improving consequential decision making under imperfect predictions [paper]
  NK, Manuel Gomez-Rodriguez, Bernhard Schölkopf, Krikamol Muandet, Isabel Valera
  KDD 2019 Workshop on Data Collection, Curation, and Labeling for Mining and Learning (DCCL)

• Generalization in anti-causal learning [paper]
  NK*,  Giambattista Parascandolo*, Bernhard Schölkopf (* equal contribution)
  NeurIPS 2018 Workshop on Critiquing and correcting trends in machine learning

• Blind Justice: Fairness with Encrypted Sensitive Attributes [paper] [bibtex] [poster] [code]
  NK, Adrià Gascón, Matt J. Kusner,  Michael Veale, Krishna P. Gummadi, Adrian Weller
  ICML 2018

• Learning Independent Causal Mechanisms [paper] [bibtex]
  Giambattista Parascandolo, NK, Mateo Rojas-Carulla, Bernhard Schölkopf
  ICML 2018

• Avoiding Discrimination Through Causal Reasoning [paper] [bibtex] [poster]
  NK, Mateo Rojas-Carulla, Giambattista Parascandolo, Moritz Hardt, Dominik Janzing, Bernhard Schölkopf
  NeurIPS 2017

• ConvWave: Searching for Gravitational Waves with Fully Convolutional Neural Nets [paper] [bibtex] [code] [poster]
  Timothy Gebhard*, NK*, Giambattista Parascandolo, Ian Harry, Bernhard Schölkopf (* equal contribution)
  NeurIPS 2017 Workshop on Deep Learning for Physical Sciences

• Universal Hydrodynamic Flow in Holographic Planar Shock Collisions [paper, arxiv version] [detailed project report]
  Paul Chesler, NK, Wilke van der Schee
  Journal for High Energy Physics, 2015

PhD thesis

Beyond traditional assumptions in fair machine learning [thesis pdf (arxiv)]
PhD Thesis @ University of Cambridge

Book

Quod erat knobelandum [springer] [amazon]
Clara Löh, Stefan Krauss, NK
Springer Spektrum (1st edition: 2016, 2nd edition: 2019)

Theses and projects

• Master Thesis Physics: Numerical Analysis of Gravitational Wave Generation during Metric Preheating [thesis] [code]

• Master Thesis Mathematics: Numerical Analysis of Causal Fermion Systems on R x S^3 [thesis]

• Physics Research Project: Sky-MoCa: The Skyrmion Phase in 3D Lattice Simulations [report] [code]

Activities

• Action Editor @TMLR

• regular area chair and reviewer at ML conferences

• Co-organizing a workshop on  Scaling up intervention models @ICML2025

• Co-organizing a workshop on A causal view on dynamical systems @NeurIPS2022

• Co-organized a workshop on Machine Learning meets Econometrics @NeurIPS2021

• Co-organized a workshop on the Neglected Assumptions in Causal Inference @ CML2021

• Co-organized a workshop on Consequential Decisions in Dynamic Environments @NeurIPS2020

• Co-organized the Human Centric Machine Learning workshop @NeurIPS2019

• Co-organized the Privacy Preserving Machine Learning workshop @NeurIPS2018

• Organized the CamTue workshop on Tenerife in 2018 and on Mallorca in 2017