World Models

Learning World Models from Physiological Time Series
Build and stress-test a self-supervised world model for biosignals, and work out where its anti-collapse regulariser stops respecting the fact that the data has an order.
Last updated on 31 Jul 2026 MSc Thesis
Forgis Research Night – Learning to Dream in EEG
Lightning talk on LuMamba — using LeJEPA “world-model” self-supervision to pretrain a tiny, montage-agnostic foundation model for biosignals.
24 Jun 2026 16:00 DARE Campus (The JED), Zürich-Schlieren, Switzerland
Code Slides LUNA paper (arXiv)
Forgis Research Night – Learning to Dream in EEG