Portrait of Thomas Möllenhoff

Thomas Möllenhoff

Deputy Team Director · Senior Research Scientist

Adaptive Bayesian Intelligence Team
RIKEN Center for Advanced Intelligence Project (AIP), Tokyo, Japan

About

I am an indefinite-term (tenured) Senior Research Scientist at RIKEN AIP and the deputy team director of the Adaptive Bayesian Intelligence Team. I am also a core member of the Bayes-Duality project. Before, I was a postdoctoral researcher with Emtiyaz Khan. I completed my PhD in the Computer Vision Group at TU Munich with Daniel Cremers.

My research aims to advance both the theory and practice of deep learning. I am currently interested in optimization and variational Bayesian principles to achieve this goal.

News

Selected Publications

  1. SOAP-Bubbles: Structured Weight Uncertainty for Neural Networks

    A. R. Minut, N. Daheim, M. Miani, M. E. Khan, W. Lin, T. Möllenhoff.

    Preprint, 2026. [code]

  2. Log-Normal Multiplicative Dynamics for Stable Low-Precision Deep Learning

    K. Nishida, E. M. Kıral, K. Bannai, M. E. Khan, T. Möllenhoff.

    ICML, 2026.

  3. Federated ADMM from Bayesian Duality

    T. Möllenhoff*, S. Swaroop*, F. Doshi-Velez, M. E. Khan.

    ICLR, 2026. [code]

  4. Variational Learning is Effective for Large Deep Networks

    Y. Shen*, N. Daheim*, B. Cong, P. Nickl, G. M. Marconi, C. Bazan, R. Yokota, I. Gurevych, D. Cremers, M. E. Khan, T. Möllenhoff.

    ICML, 2024. [code]

  5. SAM as an Optimal Relaxation of Bayes

    T. Möllenhoff, M. E. Khan.

    ICLR, 2023. [code]

  6. Proximal Backpropagation

    T. Frerix*, T. Möllenhoff*, M. Moeller*, D. Cremers.

    ICLR, 2018. [code]

* Equal contribution.