Class balance

DA_alg is a function that will help implement a data augmentation for both EEG sleep and EEG motor recordings in order to adjust the ratio of each present class or label.

For sleep recordings you will have to use :

• data_augment_sleep(recording, hypnogram, snr, param)

Parameters:

• recording: an .edf file of the sleep recording.

• hypnogram: an .edf file of the sleep hypnogram recording.

• snr: the requested SNR to be set for the noise, as the data augmentation is based on noise addition.

• param: can take 3 parameters ('g' for white gaussian noise/ 'p' for poisson noise/ 'c' for Chua noise (a chaotic noise)).

Returns:

• sleep_stage_1, sleep_stage_2, sleep_stage_3, sleep_stage_4, sleep_stage_rapid_eye_mvt,sleep_stage_wake : 5 numpy arrays from the 5 sleep stage classes with all arrays having the same length.

---

For motor recordings you will have to use :

• data_augment_motor(X, Y, snr, param)

Parameters:

• X: a numpy array [number_of_channels,time_values] containing the motor recordings.

• Y: a numpy array presenting the labels; ❗ note that only 3 labels are considered left and right hand movements and others ❗

• snr: the requested SNR to be set for the noise, as the data augmentation is based on noise addition.

• param: can take 3 parameters ('g' for white gaussian noise/ 'p' for poisson noise/ 'c' for Chua noise (a chaotic noise)).

Returns:

• motor_(right\left), motor_(left\right), motor_others : 3 numpy arrays from the 3 movement classes with all arrays having the same length.