Parameter transfer learning based on shallow visual geometry group network and its application in motor imagery classification

• Published in: Sheng wu yi xue gong cheng xue za zhi Vol. 39; no. 1; p. 28
• Main Authors: Xu, Dongqin, Li, Ming'ai
• Format: Journal Article
• Language: Chinese
• Published: China Sichuan Society for Biomedical Engineering 25.02.2022
• Subjects: Algorithms; Brain; Brain-Computer Interfaces; Calibration; Classification; Computer applications; Correlation coefficient; Correlation coefficients; Domains; EEG; Electroencephalography; Electroencephalography - methods; Fourier transforms; Human-computer interface; Humans; Image classification; Imagination; Implants; Learning; Machine Learning; Mathematical models; Mental task performance; Modules; Motor skill learning; Optimization; Parameters; Performance enhancement; Rehabilitation; Transfer learning; Visual discrimination learning; Transfer strategy; Visual geometry group network; Parameter transfer learning; Motor imagery; Pearson correlation coefficient
• ISSN: 1001-5515
• DOI: 10.7507/1001-5515.202108060

Transfer learning is provided with potential research value and application prospect in motor imagery electroencephalography (MI-EEG)-based brain-computer interface (BCI) rehabilitation system, and the source domain classification model and transfer strategy are the two important aspects that directly affect the performance and transfer efficiency of the target domain model. Therefore, we propose a parameter transfer learning method based on shallow visual geometry group network (PTL-sVGG). First, Pearson correlation coefficient is used to screen the subjects of the source domain, and the short-time Fourier transform is performed on the MI-EEG data of each selected subject to acquire the time-frequency spectrogram images (TFSI). Then, the architecture of VGG-16 is simplified and the block design is carried out, and the modified sVGG model is pre-trained with TFSI of source domain. Furthermore, a block-based frozen-fine-tuning transfer strategy is designed to quickly find and freeze the block with the greatest