pyriemann.transfer.TLScale¶

class pyriemann.transfer.TLScale(target_domain, final_dispersion=1.0, centered_data=False, metric='riemann')¶

Scaling for transfer learning.

For inputs in matrix manifold, it stretches the matrices from each domain around their mean so that the dispersion of the matrices of each domain is equal to one [1].

For inputs in tangent space, it scales the tangent vectors from each domain so that the mean of norms of vectors of each domain is equal to one.

Note

Using .fit() and then .transform() will give different results than .fit_transform(). In fact, .fit_transform() should be applied on the training set (target and source domains), and .transform() on the target domain of the test set.

Parameters:

target_domainstr: Domain to consider as target.
dispersionfloat, default=1.0: For inputs in manifold, target value for the dispersion of the matrices.
centered_databool, default=False: For inputs in manifold, whether the matrices have been re-centered to the identity matrix beforehand.
metricstr, default=”riemann”: For inputs in manifold, metric used for calculating the dispersion. For the list of supported metrics, see pyriemann.utils.distance.distance(). The stretching operation in manifold is properly defined only for the “riemann” metric.

Attributes:

scales_dict: Dictionary with key=domain_name and value=domain_scale.

See also

TLCenter

Notes

Added in version 0.4.

Changed in version 0.8: Add support for tangent space scaling.

References

[1]

Riemannian Procrustes analysis: transfer learning for brain-computer interfaces PLC Rodrigues et al, IEEE Transactions on Biomedical Engineering, vol. 66, no. 8, pp. 2390-2401, December, 2018

__init__(target_domain, final_dispersion=1.0, centered_data=False, metric='riemann')¶: Init

property dispersions_¶: Warning

DEPRECATED: Attribute dispersions_ is deprecated and will be removed in 0.10.0; please use scales_.

fit(X, y_enc, sample_weight=None)¶

Fit TLScale.

For each domain, it calculates the scaling of this domain, ie the dispersion around the mean of matrices, or the mean of the norm of vectors.

Parameters:

Xndarray, shape (n_matrices, n_channels, n_channels) or shape (n_vectors, n_ts): Set of SPD matrices or tangent vectors.
y_encndarray, shape (n_matrices,) or shape (n_vectors,): Extended labels for each matrix or vector.
sample_weightNone | ndarray, shape (n_matrices,) or shape (n_vectors,), default=None: Weights for each matrix or vector. If None, it uses equal weights.

Returns:

selfTLScale instance: The TLScale instance.

fit_transform(X, y_enc, sample_weight=None)¶

Fit TLScale and then scale each domain.

For each domain, it calculates the dispersion around the mean of this domain, and then stretches them to the desired final dispersion. For vectors, it scales them so that the mean of norms of vectors of each domain is equal to one.

Note

This method is designed for using at training time. The output for .fit_transform() will be different than using .fit() and .transform() separately.

Parameters:

Xndarray, shape (n_matrices, n_channels, n_channels) or shape (n_vectors, n_ts): Set of SPD matrices or tangent vectors.
y_encndarray, shape (n_matrices,) or shape (n_vectors,): Extended labels for each matrix or vector.
sample_weightNone | ndarray, shape (n_matrices,) or shape (n_vectors,), default=None: Weights for each matrix or vector. If None, it uses equal weights.

Returns:

X_newndarray, shape (n_matrices, n_channels, n_channels) or shape (n_vectors, n_ts): Set of scaled SPD matrices or tangent vectors in each domain.

get_metadata_routing()¶

Get metadata routing of this object.

Please check User Guide on how the routing mechanism works.

Returns:

routingMetadataRequest: A MetadataRequest encapsulating routing information.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters:

deepbool, default=True: If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:

paramsdict: Parameter names mapped to their values.

set_fit_request(*, sample_weight: bool | None | str = '$UNCHANGED$', y_enc: bool | None | str = '$UNCHANGED$') → TLScale¶

Request metadata passed to the fit method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

True: metadata is requested, and passed to fit if provided. The request is ignored if metadata is not provided.
False: metadata is not requested and the meta-estimator will not pass it to fit.
None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.
str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters:

sample_weightstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED: Metadata routing for sample_weight parameter in fit.
y_encstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED: Metadata routing for y_enc parameter in fit.

Returns:

selfobject: The updated object.

set_output(*, transform=None)¶

Set output container.

See Introducing the set_output API for an example on how to use the API.

Parameters:

transform{“default”, “pandas”, “polars”}, default=None

Configure output of transform and fit_transform.

“default”: Default output format of a transformer
“pandas”: DataFrame output
“polars”: Polars output
None: Transform configuration is unchanged

Added in version 1.4: “polars” option was added.

Returns:

selfestimator instance: Estimator instance.

set_params(**params)¶

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters:

**paramsdict: Estimator parameters.

Returns:

selfestimator instance: Estimator instance.

transform(X)¶

Scale in the target domain.

Note

This method is designed for using at test time, scaling all inputs in target domain.

Parameters:

Xndarray, shape (n_matrices, n_channels, n_channels) or shape (n_vectors, n_ts): Set of SPD matrices or tangent vectors.

Returns:

X_newndarray, shape (n_matrices, n_classes) or shape (n_vectors, n_ts): Set of scaled SPD matrices or tangent vectors in target domain.

Examples using `pyriemann.transfer.TLScale`¶

Comparison of pipelines for transfer learning

pyriemann.transfer.TLScale¶

Examples using pyriemann.transfer.TLScale¶

Examples using `pyriemann.transfer.TLScale`¶