EEG-DG: A Multi-Source Domain Generalization Framework for Motor Imagery EEG Classification

Xiao-Cong Zhong,Qisong Wang,Dan Liu,Zhihuang Chen,Jing-Xiao Liao,Jinwei Sun,Yudong Zhang,Feng-Lei Fan

2023-11-09

Abstract:Motor imagery EEG classification plays a crucial role in non-invasive Brain-Computer Interface (BCI) research. However, the classification is affected by the non-stationarity and individual variations of EEG signals. Simply pooling EEG data with different statistical distributions to train a classification model can severely degrade the generalization performance. To address this issue, the existing methods primarily focus on domain adaptation, which requires access to the target data during training. This is unrealistic in many EEG application scenarios. In this paper, we propose a novel multi-source domain generalization framework called EEG-DG, which leverages multiple source domains with different statistical distributions to build generalizable models on unseen target EEG data. We optimize both the marginal and conditional distributions to ensure the stability of the joint distribution across source domains and extend it to a multi-source domain generalization framework to achieve domain-invariant feature representation, thereby alleviating calibration efforts. Systematic experiments on a simulative dataset and BCI competition datasets IV-2a and IV-2b demonstrate the superiority of our proposed EEG-DG over state-of-the-art methods. Specifically, EEG-DG achieves an average classification accuracy/kappa value of 81.79%/0.7572 and 87.12%/0.7424 on datasets IV-2a and IV-2b, respectively, which even outperforms some domain adaptation methods. Our code is available at <a class="link-external link-https" href="https://github.com/XC-ZhongHIT/EEG-DG" rel="external noopener nofollow">this https URL</a> for free download and evaluation.

Signal Processing

What problem does this paper attempt to address?

The problem that this paper attempts to solve is that in brain - computer interface (BCI) research, the classification of motor - imagery electroencephalogram (EEG) is affected by individual differences and non - stationarity, resulting in a decline in classification performance. Specifically, since EEG signals collected at different time points and from different subjects usually exhibit different patterns, even under the same motor - imagery conditions, this requires long - term data collection, training, and calibration for each use start and each new subject, which severely limits the usability and scalability of BCI. Traditional machine - learning methods cannot solve this calibration problem because they usually assume that the training data and the test data have the same data distribution, which can significantly impair the generalization performance of the model. To overcome this challenge, the paper proposes a new multi - source domain generalization framework (EEG - DG), aiming to use different statistical distributions from multiple source domains to build a model with generalization ability for unseen target EEG data. The framework optimizes the marginal distribution \(P(X)\) and the conditional distribution \(P(Y|X)\), ensures the stability of the joint distribution among multiple source domains, and extends it to a multi - source domain generalization framework to achieve domain - invariant feature representation, thereby reducing the calibration workload. Experimental results show that EEG - DG performs better than existing methods on the simulated data set and the BCI competition data sets IV - 2a and IV - 2b, and even outperforms some domain - adaptation methods that require access to target data.

EEG-DG: A Multi-Source Domain Generalization Framework for Motor Imagery EEG Classification

Exploiting Asymmetric EEG Signals with EFD in Deep Learning Domain for Robust BCI

Domain-Generalized EEG Classification With Category-Oriented Feature Decorrelation and Cross-View Consistency Learning

Unsupervised and semi-supervised domain adaptation networks considering both global knowledge and prototype-based local class information for Motor Imagery Classification

A New Framework for Automatic Detection of Motor and Mental Imagery EEG Signals for Robust BCI Systems

A Dynamic Domain Adaptation Deep Learning Network for EEG-based Motor Imagery Classification

Deep Representation-Based Domain Adaptation for Nonstationary EEG Classification

EEG-Based Motor Imagery Recognition Framework via Multisubject Dynamic Transfer and Iterative Self-Training

An Extended Computer Aided Diagnosis System for Robust BCI Applications

Improved Domain Adaptation Network Based on Wasserstein Distance for Motor Imagery EEG Classification

A Domain Adaptation-Based Method for Classification of Motor Imagery EEG

A Deep Domain Adaptation Framework with Correlation Alignment for EEG-based Motor Imagery Classification.

Deep-learning-based motor imagery EEG classification by exploiting the functional connectivity of cortical source imaging

Multi-Class Motor Imagery EEG Decoding for Brain-Computer Interfaces

Priming Cross-Session Motor Imagery Classification with A Universal Deep Domain Adaptation Framework

Multiattention Adaptation Network for Motor Imagery Recognition

Classification of Motor Imagery Using Multisource Joint Transfer Learning

EEG Classification of Motor Imagery Using a Novel Deep Learning Framework

Dynamic Joint Domain Adaptation Network for Motor Imagery Classification

Shallow Inception Domain Adaptation Network for EEG-based Motor Imagery Classification

Multi-source deep domain adaptation ensemble framework for cross-dataset motor imagery EEG transfer learning