Anna Tegon

Recursive refinement unlocks a 44% relative gain in spatial context prediction accuracy for VLMs, even with minimal trainable parameters.

Recursive models achieve state-of-the-art multivariate time-series classification by integrating learned temporal tokenization with parameter-efficient, weight-tied state refinement.

Foundation models for EEG can now be 377x more efficient and handle 12x longer sequences, thanks to a novel Mamba-based architecture that also cracks the code for handling variable electrode setups.