The neighboring early-stopping method reduces computational complexity by using a uniform grid in inverse regularization and comparing only adjacent estimators, avoiding construction of the exact kernel Gram matrix

The method uses a grid that is uniform in inverse regularization and compares only adjacent estimators, reducing the number of discrepancy comparisons relative to standard all-pairs Lepskii-type procedures. Both the neighboring discrepancy and its empirical complexity term can be computed directly in the random feature space, without constructing the exact kernel Gram matrix.

http://arxiv.org/abs/2608.25513v1