脑机接口的使用一般需要较长的训练过程，而大量的训练会增加系统使用的不便性，加重用户疲劳，影响系统性能。尤其是在残疾人的应用中，训练集的获取非常困难。因此，在仅有少量或者没有训练集的情况下，如何让用户能够快速的使用脑机接口，是脑机接口进入临床应用与日常生活所亟待解决的重要问题。本项目通过研究跨个体脑信号解码方法，一方面促进对跨个体脑活动机制的理解；另一方面通过研究基于迁移学习的脑信号解码方法来解决脑机接口应用中小训练集甚至零训练的问题。项目分别考虑了跨时间迁移、跨个体迁移以及健康被试对残疾人迁移这些重要的实际问题。同时，在现有其他领域的迁移学习算法的基础上，改进和优化适用于脑机接口中脑信号迁移识别问题的算法。在进行相关跨个体脑信号解码理论研究的同时，进一步揭示脑机接口的生理和理论本质并推动其应用。

The use of a brain-computer interface system generally requires a long training process, which will greatly increase the inconvenience of using such a system, increase user fatigue and affect the performance. It is usually very difficult, especially for handicapped people, to collect a satisfactory training set. Therefore, how to enable users to quickly use a BCI system with only a little or even no training data is an important issue that must be solved urgently for both clinical and daily life application of BCIs. This project aims to promote the understanding of cross-subject brain activity mechanisms by studying cross-subject brain signal decoding approaches. On the other hand, by studying the brain signal decoding method based on transfer learning, the project solves the problems of small training set or even zero training problem. This project considered the important practical issues of cross-session, cross-subject and health-patient transfer respectively. At the same time, based on the existing transfer learning algorithms in other fields, the algorithms for decoding brain signal in BCI are improved and optimized. While carrying on the research on cross-subject brain signal decoding theory, we further reveal the physiological and theoretical nature of BCIs and promote its application.