mne_denoise.dss.SmoothingBias

class mne_denoise.dss.SmoothingBias(window: int = 10, iterations: int = 1)

Causal running-mean bias for DSS.

Parameters:

windowint, default=10

Smoothing-window length in samples.

iterationsint, default=1

Number of smoothing passes.

Notes

For 3D channel-first input, the implementation reshapes (n_channels, n_times, n_epochs) to (n_channels, -1) before smoothing, so the time and epoch axes are concatenated rather than smoothed independently. The original shape is restored on return.

apply(data: ndarray) → ndarray

Apply the causal running-mean bias.

Parameters:

datandarray, shape (n_channels, n_times) or (n_channels, n_times, n_epochs)

Channel-first data.

Returns:

ndarray

Smoothed data with the input shape.