Device Passport: Enabling Spatio-Temporal Pretrained Models to Generalize Across Input Layouts

2026-06-30

AI Summary

This work addresses the limited generalization of biosignal pretraining models caused by the scarcity of large-scale, device-specific datasets under novel electrode layouts. To overcome this challenge, the authors propose Device Passport, a novel channel embedding approach that uniquely integrates both functional activity and metadata of channels within a mixture-of-experts framework—surpassing prior methods that rely solely on one source of information. This strategy effectively enables spatiotemporal pretraining models to transfer across diverse input layouts. Empirical results demonstrate that Device Passport significantly outperforms the strongest baselines in both controlled subset transfer scenarios and real-world ear-EEG tasks, exhibiting particularly robust generalization when layout discrepancies are substantial.

Abstract

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 experts and mixture models that take each channel's functional activity and metadata as input. This contrasts with prior embedding methods, which typically use only functional information or only metadata to look up learned or fixed positional embeddings. Across controlled subset-transfer experiments and realistic transfer to ear-EEG, Device Passport is competitive overall and improves over the strongest learned baseline in the layout-transfer regimes that motivate this work. These results suggest that channel embedding design is a key consideration when reusing large-scale pretrained biosignal models on new devices.

Problem

• device layout
• cross-layout generalization
• biosignal foundation models
• channel embedding
• transfer learning

Innovation

• Device Passport
• channel embedding
• cross-layout transfer
• biosignal foundation models
• mixture of experts