Public datasets for emotion recognition of electroencephalogram (EEG) data and corresponding frequency domain features
Description
This dataset contains the DEAP and SEED datasets for emotion recognition of electroencephalogram (EEG) signals, as well as the frequency-domain DE features extracted by ourselves.
Files
Steps to reproduce
For the DEAP data, after eliminating the baseline signal and the last eight channels from the original 32 subjects' data, a ten-second time window was applied with an overlap rate of 50% for data augmentation, resulting in 14,080 samples. For the SEED dataset, the three experimental data of each subject are combined and data reconstruction is carried out to obtain 675 samples. After processing the raw data, the Fourier transform method is applied to convert the data to the frequency domain, and then the differential entropy features are extracted(DEAP for 5 bands, SEED for 3 bands). For different datasets or dimensions, the corresponding label information is extracted.
Institutions
- Xi'an Polytechnic University