arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2309.13993cs.LGcs.DSeess.SPstat.ML
收藏

以接近最优的样本与时间复杂度识别离散乘积分布混合

Identification of Mixtures of Discrete Product Distributions in Near-Optimal Sample and Time Complexity

California Institute of Technology(加州理工学院);The Hebrew University of Jerusalem(耶路撒冷希伯来大学)

Spencer L. Gordon, Erik Jahn, Bijan Mazaheri, Yuval Rabani, Leonard J. Schulman

AI总结:

针对 $k$ 个离散乘积分布混合的识别问题,本文结合稳健张量分解与 Hadamard 扩展矩阵条件数新界,在 $n\geq2k-1$ 时达到 $(1/ζ)^{O(k)}$ 的近优样本与时间复杂度。

摘要:

我们研究从统计信息中识别离散随机变量 $X_1,\ldots,X_n$ 的分布这一问题,该分布是 $k$ 个乘积分布的混合。对于 $n\in O(k)$,此前最优的样本复杂度为 $(1/ζ)^{O(k^2\log k)}$(在由 $ζ$ 参数化的温和可分性假设下),而已知最优下界为 $\exp(Ω(k))$。已知 $n\geq 2k-1$ 对识别而言既是必要的也是充分的。我们证明,对于任意 $n\geq 2k-1$,可实现 $(1/ζ)^{O(k)}$ 的样本复杂度和运行时间复杂度。我们还将已知的 $e^{Ω(k)}$ 下界加以扩展,使其在很宽的 $ζ$ 范围内与我们的上界匹配。这些结果通过结合以下两点获得:(a) 用于稳健张量分解的经典方法;(b) 通过研究称为 Hadamard 扩展的关键矩阵仅作用于展平的秩-1张量,以新方式界定其条件数。

英文摘要:

We consider the problem of identifying, from statistics, a distribution of discrete random variables $X_1,\ldots,X_n$ that is a mixture of $k$ product distributions. The best previous sample complexity for $n \in O(k)$ was $(1/ζ)^{O(k^2 \log k)}$ (under a mild separation assumption parameterized by $ζ$). The best known lower bound was $\exp(Ω(k))$. It is known that $n\geq 2k-1$ is necessary and sufficient for identification. We show, for any $n\geq 2k-1$, how to achieve sample complexity and run-time complexity $(1/ζ)^{O(k)}$. We also extend the known lower bound of $e^{Ω(k)}$ to match our upper bound across a broad range of $ζ$. Our results are obtained by combining (a) a classic method for robust tensor decomposition, (b) a novel way of bounding the condition number of key matrices called Hadamard extensions, by studying their action only on flattened rank-1 tensors.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑