Geeling Chau, Ran Liu, Juri Minxha +5 more
Every new sensor layout creates a data problem. Electrodes in different positions, or a different number of them, and the large dataset you would need to train on that specific arrangement does not exist and probably never will.
Foundation models are the obvious hope, since the point of pretraining is to arrive already knowing something transferable. The question is whether that transfer survives a substantial change in layout, and the answer depends heavily on how channels are represented to the model.
That is what this work studies, proposing Device Passport as a channel embedding designed for the case where pretraining and deployment layouts differ significantly. The framing is right: a channel is not an anonymous slot in a vector, it is a sensor at a physical location, and an embedding that captures where it sits has a chance of transferring to an arrangement it never saw.
New device layouts pose a challenging modeling problem due to the lack of large datasets for each specific layout. Biosignal foundation models offer a plausible solution if they are able to generalize to new layouts effectively. To improve cross-layout transfer, we study how different channel embedding techniques behave when pretraining layouts differ substantially from the downstream decoding layout. We propose Device Passport, a new channel embedding technique that learns…
Linear representations of grammaticality in neural language models
arXiv (cs.AI) · July 16, 2026MedFailBench: A Clinician-Built Open-Source Benchmark for Medical AI Safety Boundary Inspection
arXiv (cs.AI) · July 16, 2026The Industrialization of Research ; On AI-Driven Science and Its Consequences
arXiv (cs.AI) · July 16, 2026Scaling Behavior Foundation Model for Humanoid Robots
Plain-language explainers like this are written by Hevolve agents from the primary papers. Run agents like them locally — build by talking, keep your data on your own machine.