Public articles linked to the same research event.
arXiv The paper studies fair diversity maximization with exact group quotas and proposes a unified local-repair framework: starting from a solution that already meets all quotas and is color-separated, it repeatedly replaces conflicting points with a repair set, preserving quotas and within-group separation while strictly decreasing the number of bad points; this yields a polynomial-time 4-approximation for any constant number of groups and a polynomial-time 2-approximation for two groups, the latter matching the optimal factor that is NP-hard to improve even in the unconstrained case.
The paper studies fair diversity maximization with exact group quotas and proposes a unified local-repair framework: starting from a solution that already meets all quotas and is color-separated, it repeatedly replaces conflicting points with a repair set, preserving quotas and within-group separation while strictly decreasing the number of bad points; this yields a polynomial-time 4-approximation for any constant number of groups and a polynomial-time 2-approximation for two groups, the latter matching the optimal factor that is NP-hard to improve even in the unconstrained case.
The paper studies fair diversity maximization with exact group quotas and proposes a unified local-repair framework: starting from a solution that already meets all quotas and is color-separated, it repeatedly replaces conflicting points with a repair set, preserving quotas and within-group separation while strictly decreasing the number of bad points; this yields a polynomial-time 4-approximation for any constant number of groups and a polynomial-time 2-approximation for two groups, the latter matching the optimal factor that is NP-hard to improve even in the unconstrained case.
The paper studies fair diversity maximization with exact group quotas and proposes a unified local-repair framework: starting from a solution that already meets all quotas and is color-separated, it repeatedly replaces conflicting points with a repair set, preserving quotas and within-group separation while strictly decreasing the number of bad points; this yields a polynomial-time 4-approximation for any constant number of groups and a polynomial-time 2-approximation for two groups, the latter matching the optimal factor that is NP-hard to improve even in the unconstrained case.