 This paper proposes a novel approach to improving the accuracy of EEG-based motor imagery classification. It leverages existing labeled data from multiple subjects to improve the performance of MI classification on a single subject. This is achieved through the use of a Wasserstein distance-based domain adaptation network, which measures the distance between source and target domains and aligns them using an adversarial learning strategy. The proposed framework was tested on two open, source datasets, BCI competition for datasets 2A and 2B, and outperformed other state-of-the-art algorithms. This article was authored by Qing Shanxi, Tai Chen, Feng Feng, and others.