EEG-Based Mental Imagery Task Adaptation via Ensemble of Weight-Decomposed Low-Rank Adapters

TL;DR

This paper tackles EEG-based mental imagery misalignment and data scarcity by proposing EDoRA, an ensemble of weight-decomposed low-rank adapters for parameter-efficient fine-tuning on an EEG Conformer backbone. By pre-training on one mental imagery dataset (speech or motor) and fine-tuning only compact adapters on another, EDoRA achieves higher accuracy than full fine-tuning and competing PEFT methods on both speech imagery and motor imagery tasks, including cross-task transfer. Key contributions include (i) first exploration of PEFT with weight-decomposed adapters for EEG imagery, (ii) development of the ensemble EDoRA, and (iii) comprehensive analysis showing robustness and parameter efficiency across two public datasets. The approach enables efficient, scalable, and privacy-preserving brain-decoding, with potential for real-time BCI applications and cross-task rehabilitation scenarios, while highlighting directions for further optimization of adapter ranks and ensemble size.

Abstract

Electroencephalography (EEG) is widely researched for neural decoding in Brain Computer Interfaces (BCIs) as it is non-invasive, portable, and economical. However, EEG signals suffer from inter- and intra-subject variability, leading to poor performance. Recent technological advancements have led to deep learning (DL) models that have achieved high performance in various fields. However, such large models are compute- and resource-intensive and are a bottleneck for real-time neural decoding. Data distribution shift can be handled with the help of domain adaptation techniques of transfer learning (fine-tuning) and adversarial training that requires model parameter updates according to the target domain. One such recent technique is Parameter-efficient fine-tuning (PEFT), which requires only a small fraction of the total trainable parameters compared to fine-tuning the whole model. Therefore, we explored PEFT methods for adapting EEG-based mental imagery tasks. We considered two mental imagery tasks: speech imagery and motor imagery, as both of these tasks are instrumental in post-stroke neuro-rehabilitation. We proposed a novel ensemble of weight-decomposed low-rank adaptation methods, EDoRA, for parameter-efficient mental imagery task adaptation through EEG signal classification. The performance of the proposed PEFT method is validated on two publicly available datasets, one speech imagery, and the other motor imagery dataset. In extensive experiments and analysis, the proposed method has performed better than full fine-tune and state-of-the-art PEFT methods for mental imagery EEG classification.

Figures (6)

• Figure 1: (a) Overview of proposed EDoRA paramter-efficient fine-tuning approach. It depicts the overall parameter updation process of EDoRA, before and after fine-tuning (b) Feature updation via EDoRA adaptation. [$\copyright$ symbol represents concatenation, $\times$ symbol represents product, $X$ represents input features, $X^{'}$ represents output features.]
• Figure 2: Framework of the proposed method. Two experiments are performed in this work, and in these experiments EEG Conformer model is pre-trained on one dataset, and then fine-tuned on other dataset with only EDoRA adapter on each operation of transformer encoder of EEG Conformer and vice-versa. [Freezed weights are shown with lock]
• Figure 3: Mean Kappa measure of proposed method and compared methods.
• Figure 4: Confusion matrices of proposed method on two subjects of SI and MI dataset.
• Figure 5: t-SNE plots of proposed method (EDoRA) on two subjects of SI and MI dataset.