Abstract
We study approximation algorithms for the socially fair $(\ell_p,k)$-clustering problem with $m$ groups, whose special cases include thesocially fair $k$-median ($p=1$) and socially fair $k$-means ($p=2$) problems.We present (1) a polynomial-time $(5+2\sqrt{6})^p$-approximation with at most$k+m$ centers (2) a $(5+2\sqrt{6}+\epsilon)^p$-approximation with $k$ centersin time $n^{2^{O(p)}\cdot m^2}$, and (3) a $(15+6\sqrt{6})^p$ approximationwith $k$ centers in time $k^{m}\cdot\text{poly}(n)$. The first result isobtained via a refinement of the iterative rounding method using a sequence oflinear programs. The latter two results are obtained by converting a solutionwith up to $k+m$ centers to one with $k$ centers using sparsification methodsfor (2) and via an exhaustive search for (3). We also compare the performanceof our algorithms with existing bicriteria algorithms as well as exactly $k$center approximation algorithms on benchmark datasets, and find that ouralgorithms also outperform existing methods in practice.