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 2402.16719cond-mat.dis-nncond-mat.stat-mechcs.ITmath.ITmath.PRmath.STstat.TH
收藏

矩阵去噪:基于低次多项式的贝叶斯最优估计器

Matrix denoising: Bayes-optimal estimators via low-degree polynomials

Guilhem Semerjian

AI总结:

本文研究矩阵去噪问题,利用自由概率理论确定低次多项式贝叶斯最优估计器,证明其随次数增大收敛于已知估计器,并给出超越正交不变先验的普适性证据。

摘要:

我们考虑矩阵去噪问题的加性版本,其中需要从观测 $Y=S+Z$ 中推断一个大小为 $n$ 的随机对称矩阵 $S$,$Z$ 是独立随机矩阵,用于建模噪声。对于在正交矩阵共轭下不变的 $S$ 和 $Z$ 的先验分布,我们利用一阶和二阶自由概率理论的结果,在 $n$ 渐近意义下确定次数至多为 $D$ 的贝叶斯最优(就均方误差而言)多项式估计器,并证明随着 $D$ 增大,它们收敛于 Bun、Allez、Bouchaud 和 Potters 在 [IEEE Transactions on Information Theory 62, 7475 (2016)] 中引入的估计器。我们猜想这种最优性在严格正交不变先验之外仍然成立,并在 $S$ 为任意 Wishart 矩阵且 $Z$ 来自高斯正交系综(Gaussian Orthogonal Ensemble)的情形下提供了这一普适性现象的部分证据,该情形由相关的扩展秩矩阵分解问题所激发。

英文摘要:

We consider the additive version of the matrix denoising problem, where a random symmetric matrix $S$ of size $n$ has to be inferred from the observation of $Y=S+Z$, with $Z$ an independent random matrix modeling a noise. For prior distributions of $S$ and $Z$ that are invariant under conjugation by orthogonal matrices we determine, using results from first and second order free probability theory, the Bayes-optimal (in terms of the mean square error) polynomial estimators of degree at most $D$, asymptotically in $n$, and show that as $D$ increases they converge towards the estimator introduced by Bun, Allez, Bouchaud and Potters in [IEEE Transactions on Information Theory 62, 7475 (2016)]. We conjecture that this optimality holds beyond strictly orthogonally invariant priors, and provide partial evidences of this universality phenomenon when $S$ is an arbitrary Wishart matrix and $Z$ is drawn from the Gaussian Orthogonal Ensemble, a case motivated by the related extensive rank matrix factorization problem.

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。

↑