Klas Wijk

PhD Student at KTH Royal Institute of Technology

photo.png

Lindstedtsvägen 24

Stockholm, Sweden

About

I am a PhD student in Computer Science at KTH Royal Institute of Technology, supervised by Hossein Azizpour, and co-supervised by Ricardo Vinuesa at the University of Michigan. My expected graduation date is in late 2027. Before starting my PhD, I received a BSc in Computer Science and an MSc in Applied and Computational Mathematics at KTH.

My research is focused on different aspects of top-k: relaxations, sampling, and gradient estimation. I have also worked on generative models and inverse problems in fluid mechanics as part of a collaborative project. More generally, I am interested in all things machine learning, statistics, and applied mathematics.

My PhD is funded by the Swedish e-Science Research Centre (SeRC). I am also part of the Wallenberg AI, Autonomous Systems and Software Program (WASP), the largest research program in Sweden.

News

Jan 14, 2026 Attended the yearly WASP Winter Conference and presented a poster.
Oct 31, 2025 Poster accepted to the EurIPS’25 DiffSys Workshop in Copenhagen! 🇩🇰
Oct 14, 2025 Poster accepted to the ELLIS UnConference in Copenhagen! 🇩🇰
Apr 28, 2025 Attended ICLR 2025 in Singapore and presented a poster! 🇸🇬
Jan 22, 2025 Paper accepted to ICLR 2025.

Latest posts

Selected publications

  1. Differentiable Top-k: From One-Hot to k-Hot
    Klas Wijk, Ricardo Vinuesa, and Hossein Azizpour
    EurIPS 2025 Workshop on Differentiable Systems and Scientific Machine Learning, 2025
  2. SFESS: Score Function Estimators for k-Subset Sampling
    Klas Wijk, Ricardo Vinuesa, and Hossein Azizpour
    International Conference on Learning Representations, 2025
  3. Indirectly Parameterized Concrete Autoencoders
    Alfred Nilsson, Klas Wijk, Sai Gutha, and 7 more authors
    International Conference on Machine Learning, 2024

Academic Service

Reviewer for ICML 2026
Reviewer for ICLR 2026
Reviewer for EurIPS 2025 Workshop on Differentiable Systems and Scientific Machine Learning
Reviewer for ICLR 2025
Reviewer for ICML 2024 Workshop on Differentiable Almost Everything