Signal classification method and device based on fMRI high-dimensional time sequence

The invention discloses a signal classification method and device based on an fMRI (functional magnetic resonance imaging) high-dimensional time sequence, relates to the technical field of machine learning, and particularly relates to a deep learning classification algorithm which only uses functional magnetic resonance imaging data without considering any demographic information to classify subjects and does not need professionals to perform feature labeling. The method comprises the following steps of: automatically extracting features of each time step of data by using a convolutional neural network to generate a new representation, then inputting the new representation into a time sequence model Transform to learn time sequence features, and finally classifying the learned data. Compared with a traditional machine learning method, the method has the advantages that the optimal feature representation can be directly learned from complex high-dimensional data through deep learning, and a complicated and uns