Beyond Tikhonov: Faster Learning with Self-Concordant Losses via Iterative Regularization

Open in new window