Runkai Yang

We propose a new error metric for constructing coresets in \$(k,z)\$-clustering with noisy data, leading to smaller coresets, stronger theoretical guarantees, and improved empirical performance compared to classical methods.