Signal decomposition using the Spatio-Spectral Decomposition (SSD). SSD seeks to maximize the power at a frequency band of interest while simultaneously minimizing it at the flanking (surrounding) frequency bins (considered noise). It extremizes the covariance matrices associated w
| 23 | |
| 24 | @fill_doc |
| 25 | class SSD(_GEDTransformer): |
| 26 | """ |
| 27 | Signal decomposition using the Spatio-Spectral Decomposition (SSD). |
| 28 | |
| 29 | SSD seeks to maximize the power at a frequency band of interest while |
| 30 | simultaneously minimizing it at the flanking (surrounding) frequency bins |
| 31 | (considered noise). It extremizes the covariance matrices associated with |
| 32 | signal and noise :footcite:`NikulinEtAl2011`. |
| 33 | |
| 34 | SSD can either be used as a dimensionality reduction method or a |
| 35 | ‘denoised’ low rank factorization method :footcite:`HaufeEtAl2014b`. |
| 36 | |
| 37 | Parameters |
| 38 | ---------- |
| 39 | %(info_not_none)s Must match the input data. |
| 40 | filt_params_signal : dict |
| 41 | Filtering for the frequencies of interest. |
| 42 | filt_params_noise : dict |
| 43 | Filtering for the frequencies of non-interest. |
| 44 | reg : float | str | None (default) |
| 45 | Which covariance estimator to use. |
| 46 | If not None (same as 'empirical'), allow regularization for covariance |
| 47 | estimation. If float, shrinkage is used (0 <= shrinkage <= 1). For str |
| 48 | options, reg will be passed to method :func:`mne.compute_covariance`. |
| 49 | n_components : int | None (default None) |
| 50 | The number of components to extract from the signal. |
| 51 | If None, the number of components equal to the rank of the data are |
| 52 | returned (see ``rank``). |
| 53 | picks : array of int | None (default None) |
| 54 | The indices of good channels. |
| 55 | sort_by_spectral_ratio : bool (default True) |
| 56 | If set to True, the components are sorted according to the spectral |
| 57 | ratio. |
| 58 | See Eq. (24) in :footcite:`NikulinEtAl2011`. |
| 59 | return_filtered : bool (default False) |
| 60 | If return_filtered is True, data is bandpassed and projected onto the |
| 61 | SSD components. |
| 62 | n_fft : int (default None) |
| 63 | If sort_by_spectral_ratio is set to True, then the SSD sources will be |
| 64 | sorted according to their spectral ratio which is calculated based on |
| 65 | :func:`mne.time_frequency.psd_array_welch`. The n_fft parameter sets the |
| 66 | length of FFT used. The default (None) will use 1 second of data. |
| 67 | See :func:`mne.time_frequency.psd_array_welch` for more information. |
| 68 | cov_method_params : dict | None (default None) |
| 69 | As in :class:`mne.decoding.SPoC` |
| 70 | The default is None. |
| 71 | restr_type : "restricting" | "whitening" | None |
| 72 | Restricting transformation for covariance matrices before performing |
| 73 | generalized eigendecomposition. |
| 74 | If "restricting" only restriction to the principal subspace of signal_cov |
| 75 | will be performed. |
| 76 | If "whitening", covariance matrices will be additionally rescaled according |
| 77 | to the whitening for the signal_cov. |
| 78 | If None, no restriction will be applied. Defaults to "whitening". |
| 79 | |
| 80 | .. versionadded:: 1.11 |
| 81 | rank : None | dict | ‘info’ | ‘full’ |
| 82 | As in :class:`mne.decoding.SPoC` |
no outgoing calls