mne_denoise.dss.TanhMaskDenoiser

class mne_denoise.dss.TanhMaskDenoiser(alpha: float = 1.0, *, normalize: bool = True)

Scaled hyperbolic-tangent nonlinearity.

The transform is tanh(alpha * source). When normalize=True, the source is divided by its standard deviation before the transform and the result is rescaled by the same value.

Parameters:

alphafloat, default=1.0

Tanh scale.

normalizebool, default=True

Whether to standardize and rescale the source around the transform.

denoise(source: ndarray) → ndarray

Apply the scaled hyperbolic-tangent nonlinearity.

Parameters:

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

Source time series. Normalization, when enabled, is computed over the supplied values.

Returns:

denoisedndarray, same shape as source

tanh(alpha * source) with optional standard-deviation scaling.