Frequency-Decorrelated Temporal Ensembles for EEG--fNIRS Imagined-Handwriting Decoding

2026-08-04

AI Summary

This study addresses the challenge of low recognition accuracy in cross-subject decoding of imagined handwriting from electroencephalography (EEG) signals, which is primarily hindered by high noise levels and substantial inter-subject variability in writing trajectories. To overcome this, the work proposes a multi-scale temporal modeling approach that treats imagined handwriting as a multi-second motor sequence and trains parallel models across three complementary EEG frequency bands. A frequency-decorrelated multi-view ensemble strategy is introduced to enhance generalization. Furthermore, the framework integrates a protocol-aware Hungarian assignment decoding mechanism, pseudo-label transfer learning, and structured posterior aggregation, collectively yielding significant improvements in cross-subject performance. The method achieves accuracies of 0.8498, 0.7718, and 0.7952 on public, private, and combined test sets, respectively, underscoring the critical roles of frequency diversity and protocol-matched inference, while revealing limited contribution from fNIRS data.

Abstract

Imagined handwriting offers a temporally rich paradigm for non-invasive neural decoding, yet reliable recognition across unseen participants remains difficult because scalp EEG is noisy and internally generated stroke sequences vary across individuals. The Multimodal Brain-Computer Interface Grand Challenge provides synchronized EEG and fNIRS for four-class subject-independent handwriting-trajectory classification. We propose FRED, a task-adapted system that models imagined handwriting as a multi-second motor sequence and trains a compact multi-scale temporal network on three complementary EEG frequency views. With three seeds per view, cross-band members produce substantially less-correlated errors than same-band replicas, yielding a clean nine-member ensemble accuracy of 0.8076/0.7242/0.7492 on the public/private/overall test partitions without test-set adaptation or output constraints. The submitted pipeline further incorporates transductive pseudo-label training, three EEG-Conformer members, posterior aggregation, and a paradigm-aware decoder. Because every 12-trial randomization block contains three instances of each class, the final predictions are obtained by Hungarian assignment under the known block quota. On one fixed posterior pool, independent, session-constrained, and block-constrained decoding achieve 0.7600, 0.7758, and 0.7952 overall accuracy, respectively. The complete system reaches 0.8498/0.7718/0.7952, ranking fourth on the private split. A modality audit finds fNIRS-only decoding at chance (0.2511 overall), while adding fNIRS to EEG changes accuracy by only +0.0025. These results identify frequency-diverse temporal EEG modeling and protocol-matched structured inference as the principal sources of performance in this sparse-montage EEG--fNIRS setting. The source code is available at https://github.com/XiuFan719/EEG-fNIRS-fuse-method-for-MM-challenge.

Problem

• imagined handwriting
• subject-independent decoding
• EEG-fNIRS
• brain-computer interface
• neural decoding

Innovation

• frequency-decorrelated ensemble
• multi-scale temporal network
• structured inference
• subject-independent decoding
• EEG-fNIRS multimodal BCI