跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2609.15677 · v3

Beyond 1/3-Approximation: Random Greedy for Constrained \(\boldsymbol{k}\)-Submodular Maximization

首次提交 2026-09-14 22:52
版本更新 2026-11-19 13:37
雷达观测 2026-11-20 19:58
arXiv ↗PDF ↗

摘要

kk-submodularity generalizes submodularity by allowing each selected element to be assigned one of kk labels. We study the maximization of a non-monotone kk-submodular function, where k≥2k\ge2, under matroid and knapsack constraints. We develop a proportional Top-2 randomized framework that achieves an approximation ratio of 2−1≈0.4142\sqrt2-1\approx0.4142 under either constraint, breaking the previous 1/31/3 barrier. Once an element is selected, its label is chosen from the two labels with the largest marginal gains, with probabilities proportional to their positive parts. We further improve the approximation guarantees to 1/21/2 under cardinality and knapsack constraints, through a Bellman-style backward induction analysis using the same proportional Top-2 rule. Our 1/21/2 guarantees for cardinality and knapsack constraints are asymptotically tight as k→∞k\to\infty. They are also tight for k=2k=2. The value-oracle complexities are O(Bnk)O(Bnk) for cardinality, where BB is the cardinality bound, O(n2k)O(n^2k) for matroid, and O(n3k2)O(n^3k^2) for knapsack.

作者备注:40 pages

AI 论文解读

基于摘要 · 非全文审阅

此版本尚未生成 AI 解读。解读由后台任务生成并按论文版本保存,登录后共享已有结果。

版本差异

本站首次收录为 v3,暂无更早版本可对照。

观测到的版本

相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。

v3 · 2026-11-19 13:37 · 最新观测版本

Beyond 1/3-Approximation: Random Greedy for Constrained \(\boldsymbol{k}\)-Submodular Maximization

kk-submodularity generalizes submodularity by allowing each selected element to be assigned one of kk labels. We study the maximization of a non-monotone kk-submodular function, where k≥2k\ge2, under matroid and knapsack constraints. We develop a proportional Top-2 randomized framework that achieves an approximation ratio of 2−1≈0.4142\sqrt2-1\approx0.4142 under either constraint, breaking the previous 1/31/3 barrier. Once an element is selected, its label is chosen from the two labels with the largest marginal gains, with probabilities proportional to their positive parts. We further improve the approximation guarantees to 1/21/2 under cardinality and knapsack constraints, through a Bellman-style backward induction analysis using the same proportional Top-2 rule. Our 1/21/2 guarantees for cardinality and knapsack constraints are asymptotically tight as k→∞k\to\infty. They are also tight for k=2k=2. The value-oracle complexities are O(Bnk)O(Bnk) for cardinality, where BB is the cardinality bound, O(n2k)O(n^2k) for matroid, and O(n3k2)O(n^3k^2) for knapsack.

查看此版本解读 →查看此版本原文 ↗