mne_denoise.dss.KurtosisDenoiser#

class mne_denoise.dss.KurtosisDenoiser(nonlinearity: str = 'tanh', alpha: float = 1.0)[source]#

Configurable ICA contrast nonlinearity.

Parameters:
nonlinearity{“tanh”, “cube”, “gauss”}, default=”tanh”

Transform to apply: tanh(alpha * source), source**3, or source * exp(-0.5 * (alpha * source)**2).

alphafloat, default=1.0

Scale used by the "tanh" and "gauss" transforms.

denoise(source: ndarray) ndarray[source]#

Apply the configured ICA contrast nonlinearity.

Parameters:
sourcendarray, shape (n_times,) or (n_times, n_epochs)

Source time series.

Returns:
denoisedndarray, same shape as source

Transformed source using tanh, cube, or gauss.