Search Results

Clustering is a fundamental problem of spatio-temporal data analysis. Given a set 𝒳 of n moving entities, each of which corresponds to a sequence of τ time-stamped points in ℝ^d, a k-clustering of 𝒳 is a partition of 𝒳 into k disjoint subsets that optimizes a given objective function. In this paper, we consider two clustering problems, k-Center and k-MM, where the goal is to minimize the maximum value of the objective function over the duration of motion for the worst-case input 𝒳. We show that both problems are NP-hard when k is an arbitrary input parameter, even when the motion is restricted to ℝ. We provide an exact algorithm for the 2-MM clustering problem in ℝ^d that runs in O(τ d n²) time. The running time can be improved to O(τ n log{n}) when the motion is restricted to ℝ. We show that the 2-Center clustering problem is NP-hard in ℝ². Our 2-MM clustering algorithm provides a 1.15-approximate solution to the 2-Center clustering problem in ℝ². Moreover, finding a (1.15-ε)-approximate solution remains NP-hard for any ε >0. For both the k-MM and k-Center clustering problems in ℝ^d, we provide a 2-approximation algorithm that runs in O(τ d n k) time.