An Enhanced Source-Free Unsupervised Domain Adaptation Framework for Cross-Dataset EEG Emotion Recognition via Predictive Coding and Test-Time Training
Md Niaz Imtiaz, Naimul Khan · 2026
Read An Enhanced Source-Free Unsupervised Domain Adaptation Framework for Cross-Dataset EEG Emotion Recognition via Predictive Coding and Test-Time Training by Md Niaz Imtiaz, Naimul Khan (2026) 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-2606-28202.txt.
Description
Md Niaz Imtiaz and Naimul Khan aim to recognize emotions from brain-wave recordings. The task breaks down when a system trained on one group of people or one dataset is used on another. Privacy rules often block sharing the original training data, so adaptation must work without it.
Imtiaz and Khan combine three ideas. The first is self-supervised pretraining that learns by predicting brain signals over time. The second is a two-stage adaptation that clusters the new data while tolerating noisy guessed labels. The third is light retraining at test time for uncertain cases. Across the DEAP, SEED, and DREAMER datasets, their method beat rival source-free approaches. Accuracy reached about seventy percent on SEED and sixty-three percent on DREAMER when starting from DEAP.
Catalog record
| CREATOR | Md Niaz Imtiaz, Naimul Khan |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 44 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.28202