Adversarially Robust Streaming Algorithms for $(k, z)$-Clustering
We design adversarially robust algorithms for dynamic geometric streams that $(1 + \epsilon)$-approximate $k$-Median and $k$-Means in Euclidean space $\mathbb{R}^d$, using $\tilde{O}(kd \mathrm{poly}(\epsilon^{-1}))$ space. This result significantly improves a recent $O(d\log n)$-approximation for $k$-Median [Gribelyuk...