Uppsala Learning, Inference and Optimization Lab – Scientific Machine Learning: AI and Computational Science ⇔ Science and Engineering

Mission

We study and advance the foundations of data-driven science and engineering.

We power scientific discovery, driven by AI, Optimization and Statistics.

We are a group hosted at the Division of Scientific Computing, Department of Information Technology, Science for Life Laboratory (SciLifeLab), UU.

We work on the fundamentals of machine learning and AI, with a strong focus on solving hard problems in (life) sciences.

News

Sep, 2026: Open PostDoc positions: (i) simulation-based inference for particle physics, (ii) probabilistic methods for foundation and world models
May, 2026: Paper on epistemic UQ for VLMs via Riemannian Flow Matching accepted at ICML 2026 – arXiv link
Jan, 2026: Our paper on conditional diffusion models for simulation-based inference accepted at AISTATS 2026 – arXiv link
Nov, 2025: Paper just out in BMC Genomics: ‘Predicting allergy and postpartum depression from an incomplete compositional microbiome‘ – link
Sep, 2025: Our paper exploring asymmetric uncertainty structures in pre-trained VLMs accepted at NeurIPS 2025 – arXiv link
May, 2025: We will participate in the NEST project ‘Time-Resolved Imaging and Multi-Channel Evaluation of Cellular Dynamics’.
July 23-25, 2024: Aleksandr Karakulev will present our take on parameter-free robust learning via variational inference at ICML 2024 – arXiv link.

Spotlight: Jan 2024

Recent student projects:

– Bayesian Sequential Model Optimization for Drug Combination Repurposing (Dhanushki Mapitigama, Mina Badri, Ema Duljkovic)

– Bayesian Optimization for Characterising Quantum Entanglement (Stefanos Tsampanakis, Ramin Modaresi, Niklas Kostrzewa)

– Deep Learning for Ill-Posed Inverse Problems in Photonics (Johan Rensfeldt, Fredrik Gillgren, Gustav Fredrikson)

Dhanushki, Mina and Ema present our work on Bayesian active learning for efficient drug combination repurposing.

Stefanos, Ramin and Niklas present our poster on a Bayesian optimization approach towards characterising quantum entanglement accurately and efficiently.

Johan, Fredrik and Gustav present our work on deep learning approaches to solve ill-posed inverse problems in nanophotonics.

Research Areas

Our research activities broadly span theory and applications within machine learning, optimization, inverse problems, statistical sampling (Publications).