Differentiating brain states via multi-clip random fragment strategy-based interactive bidirectional recurrent neural network

Shu Zhang, Enze Shi, Lin Wu, Ruoyang Wang, Sigang Yu, Zhengliang Liu, Shaochen Xu, Tianming Liu, Shijie Zhao

Abstract: Highlights•McRFS-IBiRNN shows promising characterization ability in brain state differentiation.•The proposed McRFS suppresses erroneous learning, enhances data structure diversity.•The proposed interactive BiRNN deeply fuses forward and backward representations.•The sparse penalty algorithm helps the FC layer extract activated brain regions.