/ EN
2026-04-07 00:00 China Papers Foundations & Methods Translated from EN

Dynamic Source Domain Selection: An Adaptive EEG Transfer Learning Framework

Summary To reduce negative transfer in motor imagery BCIs, the study presents an adaptive dynamic transfer learning framework that decomposes EEG time-frequency features via wavelet convolution, matches source and target samples through a dynamic transfer attention module, and uses a joint loss to shrink marginal and class-conditional differences. On BNCI2014001, BNCI2014002 and BNCI2015001 it reached 78.78%, 82.11% and 78.19% accuracy, averaging 0.13% to 27.7% above baseline algorithms.
Why it matters Selecting source domains dynamically instead of pooling them naively targets negative transfer — the failure mode driven by low source-target similarity and inter-subject variability that undercuts cross-subject MI-BCI transfer.

This study, published in Journal of Neuroscience Methods, proposes an adaptive EEG dynamic transfer learning framework that first performs time-frequency decomposition using wavelet transform convolution, then uses a dynamic migration-based attention module to match source and target domain samples in a latent space, mitigating negative transfer in motor imagery BCIs.

Validated on the BNCI2014001, BNCI2014002 and BNCI2015001 datasets, the accuracy rates were 78.78%, 82.11% and 78.19% respectively, with average accuracy outperforming baseline algorithms by 0.13% to 27.7%, and the authors say the method is robust to subject variability. PMID 41941946, DOI 10.1016/j.jneumeth.2026.110768

Compiled by BCIwiki from public sources

Sources · 1

Guangzhou University timeline

2026-09 48-Subject Study Shows 3-Channel P300 Matches 8-Channel Performance All entries →
Read original ↗
Suggest a correction Revisions · none / EN
© 2026 BCIwiki.com Digest Topics Tips Subscribe Revisions About