Transfer Learning With Optimal Transportation and Frequency Mixup for EEG-Based Motor Imagery Recognition

Objective. In the field of motor imagery (MI) electroencephalography (EEG) based brain-computer interfaces (BCIs), deep transfer learning (TL) has proven to be an effective tool for solving the problem of limited availability of subject-specific data for training of robust deep learning (DL) models. Despite considerable progress has been made in cross-subject/session and cross-device scenarios, the more challenging problem of cross-task deep TL still remains largely unexplored. Approach. We propose a novel explainable cross-task adaptive TL method for MI EEG decoding. Firstly, similarity analysis and data alignment are performed for EEG data of motor execution (ME) and MI tasks. Afterward, the MI EEG decoding model is obtained via pre-training with extensive ME EEG data and fine-tuning with partial MI EEG data. Finally, expected gradients based post-hoc explainability analysis is conducted for visualization of important temporal-spatial features. Main results. Extensive experiments are conducted on one large ME EEG dataset (HGD) and two large MI EEG datasets (openBMI and GIST). The best average classification accuracy of our method reaches 80.00% and 72.73% for openBMI and GIST respectively, which outperforms several state-of-the-art algorithms. Besides, the results of explainability analysis further validate the correlation between ME and MI EEG data and effectiveness of ME/MI cross-task adaptation. Significance. This paper confirms that decoding of MI EEG can be well facilitated by pre-existing ME EEG data, which largely relaxes the constraint of training samples for MI EEG decoding and has important practical sense.

Section: Tl Methods For MI Eeg Decodingmentioning

confidence: 99%

Explainable cross-task adaptive transfer learning for motor imagery EEG classification

Miao,

Yang,

Zeng

et al. 2023

“…Transfer learning (TL) [5] refers to the idea that knowledge studied in the source domain can be applied to the target domain to improve the performance of the target domain. TL represents a promising strategy utilized in BCI to reduce calibration time by leveraging similarities between data or models learned in the source domain to facilitate learning in the target domain [6][7][8][9][10][11][12].…”

Section: Introductionmentioning

confidence: 99%

“…Initially, OT was introduced for domain adaptation [23]. Peterson et al [10] and Chen et al [11] demonstrate the effectiveness of OT in the field of EEG-based MI. In order to mitigate the impact of interfering instances, Zhao et al [24] proposed the Density Peak Landmarks (DPLs) method, which identifies density peak instances for Grassman manifolds learning.…”

Section: Introductionmentioning

confidence: 99%

Transfer learning with data alignment and optimal transport for EEG based motor imagery classification

Chu,

Zhu,

Huang

et al. 2024

The non-stationarity of electroencephalogram signals and the variability of different subjects pose major challenges in current Brain-Computer Interfaces research, which requires a time-consuming specific calibration procedure to address. Transfer Learning can make use of data or models from one or more source domains to encourage learning in the target domain, so as to address the challenges of EEG non-stationarity and inter-subject variability. In this paper, a Multi-source domain Transfer Learning Fusion (MTLF) framework is proposed to address the non-stationarity of EEG signals and the complexity of the calibration process. Firstly, the method transforms the source domain data with the resting state segment, so as to decrease the differences between the source domain and the target domain. Then, CSP is used to extract features. After that, Optimal Transport is employed to further refine the source domain data to increase the similarity of the probability distribution between the target and source domains. Finally, an improved transfer learning classifier is employed to classify the target samples. This method does not require the label information of target domain samples and can reduce the calibration workload. The proposed MTLF is assessed on Datasets IIa and IIb of the BCI Competition IV. Compared with seven other algorithms, our method performed relatively the best and achieved mean classification accuracy of 73.69% and 70.83% on Datasets IIa and IIb respectively. Experimental results demonstrate that MTLF can effectively decrease the discrepancy between the source and target domain, and acquire better classification performance on two MI-EEG datasets.

“…DJDAN employed conditional and marginal distribution adaptation to reduce the joint distribution difference between source and target domain data and improved the performance of the cross-session MI classification tasks [33]. JDAOT-Mix proposed a domain adaptation method with optimal transport and frequency mixup that improved the performance of the MI classification tasks [34]. In [35], SACNN utilized the maximum mean discrepancy to minimize the distance of representation distribution across domains.…”

Section: Introductionmentioning

confidence: 99%

Self-supervised contrastive learning for EEG-based cross-subject motor imagery recognition

Li,

Sun

et al. 2024

Self Cite

Objective. The extensive application of electroencephalography (EEG) in brain-computer interfaces (BCIs) can be attributed to its non-invasive nature and capability to offer high-resolution data. The acquisition of EEG signals is a straightforward process, but the datasets associated with these signals frequently exhibit data scarcity and require substantial resources for proper labeling. Furthermore, there is a significant limitation in the generalization performance of EEG models due to the substantial inter-individual variability observed in EEG signals. Approach. To address these issues, we propose a novel self-supervised contrastive learning framework for decoding motor imagery (MI) signals in cross-subject scenarios. Specifically, we design an encoder combining convolutional neural network and attention mechanism. In the contrastive learning training stage, the network undergoes training with the pretext task of data augmentation to minimize the distance between pairs of homologous transformations while simultaneously maximizing the distance between pairs of heterologous transformations. It enhances the amount of data utilized for training and improves the network’s ability to extract deep features from original signals without relying on the true labels of the data. Main results. To evaluate our framework’s efficacy, we conduct extensive experiments on three public MI datasets: BCI IV IIa, BCI IV IIb, and HGD datasets. The proposed method achieves cross-subject classification accuracies of 67.32%, 82.34%, and 81.13% on the three datasets, demonstrating superior performance compared to existing methods. Significance. Therefore, this method has great promise for improving the performance of cross-subject transfer learning in MI-based BCI systems.