BCIwiki (bciwiki.com) — A Zhejiang University of Technology team developed MOPC to update motor-imagery EEG decoders as new signals arrive, with adaptation taking about 40–60 ms per trial. The paper was uploaded to arXiv on October 21, 2026. Its record states that it was accepted by IEEE Transactions on Instrumentation and Measurement and supplies the journal article DOI.
MOPC addresses changes in EEG over time and differences between users. During decoding, it processes incoming trials in a single pass without revisiting the original training data or first collecting a complete dataset from a new user.
Randomly masking parts of a signal encourages consistent predictions under interference. Feature similarity and prediction confidence reduce the influence of unreliable predictions used as training labels. A short-term memory stores recent trials and predictions to help track changing signals and constrain updates.
The authors report gains of 2.22% and 3.67% over prior methods on BCI IV IIb and OpenBMI. The 40–60 ms figure measures adaptation per trial, not the full interval from EEG acquisition to a device response. The results are dataset evaluations; practical online use requires further testing.