← 研踪 PaperTrailarXiv:2611.25011 · v1
Near-Quadratic Sample Compression for All VC Classes
首次提交 2026-11-19 02:57
版本更新 2026-11-19 02:57
雷达观测 2026-11-21 12:18
摘要
The sample compression conjecture is a longstanding open problem in learning theory, originating in the work of Littlestone and Warmuth in the 1980s. At its heart is a connection between simplification and generalization: if a predictor consistent with a training sample is defined using only a small part of that sample, then it is likely to generalize. Sample compression schemes formalize this idea, and many natural learning algorithms can be analyzed this way. The sample compression conjecture of Floyd and Warmuth (1995) asserts that every binary concept class of VC dimension admits a sample compression scheme of size ; a weaker conjecture asks only for size polynomial in . Both have remained open: the first bound depending only on , due to Moran and Yehudayoff (2016), is exponential in . We prove the weaker conjecture by constructing a sample compression scheme of size .
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。
v1 · 2026-11-19 02:57 · 最新观测版本
Near-Quadratic Sample Compression for All VC Classes
The sample compression conjecture is a longstanding open problem in learning theory, originating in the work of Littlestone and Warmuth in the 1980s. At its heart is a connection between simplification and generalization: if a predictor consistent with a training sample is defined using only a small part of that sample, then it is likely to generalize. Sample compression schemes formalize this idea, and many natural learning algorithms can be analyzed this way. The sample compression conjecture of Floyd and Warmuth (1995) asserts that every binary concept class of VC dimension admits a sample compression scheme of size ; a weaker conjecture asks only for size polynomial in . Both have remained open: the first bound depending only on , due to Moran and Yehudayoff (2016), is exponential in . We prove the weaker conjecture by constructing a sample compression scheme of size .
查看此版本解读 →查看此版本原文 ↗