EDAPT: Towards Calibration-Free BCIs with Continual Online Adaptation
Lisa Haxel *, Jaivardhan Kapoor *, Ulf Ziemann, Jakob H. Macke · 2025
Read EDAPT: Towards Calibration-Free BCIs with Continual Online Adaptation by Lisa Haxel *, Jaivardhan Kapoor *, Ulf Ziemann, Jakob H. Macke (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-10474.txt.
Description
Lisa Haxel and colleagues introduced EDAPT in 2025, a framework meant to free brain-computer interfaces from repeated calibration. These interfaces decode neural signals to control devices, but the signals drift over time and differ between people, so accuracy fades unless the system is recalibrated, a burden that blocks everyday use. EDAPT instead keeps learning while the interface runs.
The framework first trains a starting decoder on data pooled from many users, then continually fine-tunes it for the individual as their neural patterns change during use. Tested across nine datasets covering three interface tasks, it consistently beat conventional static methods, with most of the gain coming from that pairing of population pre-training and online personalization. Updates finish within a fraction of a second on consumer hardware, and accuracy depended on the total data budget rather than how it was split.
Catalog record
| CREATOR | Lisa Haxel *, Jaivardhan Kapoor *, Ulf Ziemann, Jakob H. Macke |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 68 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-10474