Almost-Optimal Upper and Lower Bounds for Clustering in Low Dimensional Euclidean Spaces
This paper improves the running time of -approximation algorithms for -median and -means clustering in low-dimensional Euclidean spaces to $2^{\tilde{O}(1/\varepsilon)^{d-1}} \cdot n \cdot \text{polylog}(n)1/\varepsilond$ is essentially optimal.