Question about feature dimension definition.

[kaio-99]

I have read your paper “HetEmotionNet: Two-Stream Heterogeneous Graph Recurrent Neural Network for Multi-modal Emotion Recognition”, what an excellent and interesting paper. According to my understanding in section 5.2 Baseline Models and Settings, a 1s non-overlapping window is employed to divide each trial into several samples, the dimension of input signals should be 32x40x60x32x4 (subjects x trials x segments x channels x bands) for DEAP dataset, so we got 32x40x60=76800 1s-length samples with dimension of 32x4. There seems to be no dimension of time_step, how to extract the temporal feature?

[Call-me-akan]

hi,you can read the dissertation"SST-EmotionNet: Spatial-Spectral-Temporal based Attention 3D Dense Network for EEG Emotion Recognition",it also be a nice essay,and it might contain the way to gain temproal feature , please ask for have you already relise the code now?I mean，have you already formulate the code about preparing？