A spectral least-squares-type method for heavy-tailed corrupted regression with unknown covariance \& heterogeneous noise

Open in new window