New paper from the lab led by Fabian Mikulasch. 📄 https://arxiv.org/abs/2609.37789 🌐 https://info-ldm.github.io/ Code: https://github.com/fmi-basel/identifiable-stochastic-nuisance Why does predicting in latent space (JEPA, CPC, SimCLR…) work so well on messy data, with changing lighting, camera angles,Continue reading
Learning Commute-Time-Preserving World Models for Planning
New paper led by Michael Hauri & Peter Buttaroni: “Learning Commute-Time-Preserving World Models for Planning.” We show how to shape a world model’s latent space so its distances mean something for planning. 📄 https://arxiv.org/abs/2610.01373 💻 https://github.com/fmi-basel/commute-time-preserving-world-models 🌐Continue reading
The Lab at the Bernstein Conference
Like every year, the lab is excited to attend and present at the Bernstein Conference and Satellite workshops. Workshop talks Monday 28th, 15:05-15:35 (room 3.104) Rory will give the talk “Connectome-constrained spiking network models ofContinue reading
Paper: Temporal prediction captures retinal spiking responses across animal species
Luke’s paper “Temporal prediction captures retinal spiking responses across animal species” is now out in Nature Communications! https://www.nature.com/articles/s41467-026-77399-y The retina’s function has long been framed as one of two extremes: an efficient compressor of visualContinue reading
Paper: Understanding Self-Supervised Learning via Latent Distribution Matching
Fabian’s paper “Understanding Self-Supervised Learning via Latent Distribution Matching” was accepted as ICML spotlight! https://arxiv.org/abs/2605.03517 We unify self-supervised learning (SSL) algorithms (e.g., contrastive, VICReg, stopgrad) via latent distribution matching (LDM), which matches an induced latentContinue reading
Cosyne 2026
The lab is at Cosyne 2026 with three posters! Go and check them out if you are there in Lisbon! 2-069: Learning representations of moving objects in a reconstruction-free cortical circuit modelFriday 13th, 13:15-16:15 (AtenaContinue reading
New paper: Dreamer-CDP — Reconstruction-Free World Models for Reinforcement Learning
In our new tiny paper accepted at the ICLR workshop on world models we introduce Dreamer-CDP, a Dreamer variant that learns a world model without reconstructing raw pixel observations. Preprint: https://arxiv.org/abs/2603.07083 Standard model-based reinforcement learningContinue reading
Understanding neural circuit principles for representation learning through joint-embedding predictive architectures
We’re happy to share our new preprint “Understanding neural circuit principles for representation learning through joint-embedding predictive architectures” led by Atena and Manu 🚀 We looked into the question how the cortex learns to representContinue reading
New mates on our crew
This year we gladly welcome four new Kraken Crew members: Fabian, Thomas, Michael, and Rory (jointly with the Friedrich lab). We look forward to what we will learn from your cool research projects and toContinue reading
SNUFA 2025
We are stoked to bring to you another edition of SNUFA our friendly online workshop on spiking neural networks as universal function approximators. This year on 5-6 November 2025, European afternoons. Like every year, weContinue reading