pyriemann.transfer.MDWM

class pyriemann.transfer.MDWM(domain_tradeoff, target_domain, metric='riemann', n_jobs=1)

Classification by Minimum Distance to Weighted Mean.

Classification by nearest centroid. For each of the given classes, a centroid is estimated, according to the chosen metric, as a weighted mean of SPD matrices from the source domain, combined with the class centroid of the target domain [1] [2]. For classification, a given new matrix is attibuted to the class whose centroid is the nearest according to the chosen metric.

Parameters
domain_tradeofffloat

Coefficient in [0,1] controlling the transfer, ie the trade-off between source and target domains. At 0, there is no transfer, only matrices acquired from the source domain are used. At 1, this is a calibration-free system as no matrices are required from the source domain.

target_domainstring

Name of the target domain in extended labels.

metricstring | dict, default=’riemann’

The type of metric used for centroid and distance estimation. see mean_covariance for the list of supported metric. the metric could be a dict with two keys, mean and distance in order to pass different metric for the centroid estimation and the distance estimation. Typical usecase is to pass ‘logeuclid’ metric for the mean in order to boost the computional speed and ‘riemann’ for the distance in order to keep the good sensitivity for the classification.

n_jobsint, default=1

The number of jobs to use for the computation. This works by computing each of the class centroid in parallel. If -1 all CPUs are used. If 1 is given, no parallel computing code is used at all, which is useful for debugging. For n_jobs below -1, (n_cpus + 1 + n_jobs) are used. Thus for n_jobs = -2, all CPUs but one are used.

See also

MDM

Notes

New in version 0.4.

References

1

Transfer learning for SSVEP-based BCI using Riemannian similarities between users E. Kalunga, S. Chevallier and Q. Barthelemy, in 26th European Signal Processing Conference (EUSIPCO), pp. 1685-1689. IEEE, 2018.

2

Minimizing Subject-dependent Calibration for BCI with Riemannian Transfer Learning S. Khazem, S. Chevallier, Q. Barthelemy, K. Haroun and C. Nous, 10th International IEEE/EMBS Conference on Neural Engineering (NER), pp. 523-526. IEEE, 2021.

Attributes
classes_ndarray, shape (n_classes,)

Labels for each class.

covmeans_list of n_classes ndarrays of shape (n_channels, n_channels)

Centroids for each class.

__init__(domain_tradeoff, target_domain, metric='riemann', n_jobs=1)

Init.

fit(X, y_enc, sample_weight=None)

Fit (estimates) the centroids.

Parameters
Xndarray, shape (n_matrices, n_channels, n_channels)

Set of SPD matrices from source and target domain.

y_encndarray, shape (n_matrices,)

Extended labels for each matrix.

sample_weightNone | ndarray, shape (n_matrices_source,), default=None

Weights for each matrix from the source domains. If None, it uses equal weights.

Returns
selfMDWM instance

The MDWM instance.

fit_predict(X, y)

Fit and predict in one function.

fit_transform(X, y=None, **fit_params)

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters
Xarray-like of shape (n_samples, n_features)

Input samples.

yarray-like of shape (n_samples,) or (n_samples, n_outputs), default=None

Target values (None for unsupervised transformations).

**fit_paramsdict

Additional fit parameters.

Returns
X_newndarray array of shape (n_samples, n_features_new)

Transformed array.

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.

predict(X)

Get the predictions.

Parameters
Xndarray, shape (n_matrices, n_channels, n_channels)

Set of SPD matrices.

Returns
predndarray of int, shape (n_matrices,)

Predictions for each matrix according to the closest centroid.

predict_proba(X)

Predict proba using softmax.

Parameters
Xndarray, shape (n_matrices, n_channels, n_channels)

Set of SPD matrices.

Returns
probndarray, shape (n_matrices, n_classes)

Probabilities for each class.

score(X, y_enc, sample_weight=None)

Return the mean accuracy on the given test data and labels.

Parameters
Xndarray, shape (n_matrices, n_channels, n_channels)

Set of SPD matrices.

y_encndarray, shape (n_matrices,)

Extended labels for each matrix.

Returns
scorefloat

Mean accuracy of clf.predict(X) wrt. y_enc.

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)

Get the distance to each centroid.

Parameters
Xndarray, shape (n_matrices, n_channels, n_channels)

Set of SPD matrices.

Returns
distndarray, shape (n_matrices, n_classes)

The distance to each centroid according to the metric.