pyriemann.utils.mean.mean_logeuclid¶

pyriemann.utils.mean.mean_logeuclid(covmats, sample_weight=None)[source]¶

Return the mean covariance matrix according to the log-euclidean metric.

$$\mathbf{C} = \exp{(\frac{1}{N} \sum_i \log{\mathbf{C}_i})}$$

Parameters:	covmats – Covariance matrices set, Ntrials X Nchannels X Nchannels sample_weight – the weight of each sample
Returns:	the mean covariance matrix