arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2102.10217cs.ITmath.IT
收藏

从列线性和二次投影中快速和样本有效的联邦低秩矩阵恢复

Fast and Sample-Efficient Federated Low Rank Matrix Recovery from column-wise Linear and Quadratic Projections

Seyedehsara, Nayer, Namrata Vaswani

摘要:

我们研究了以下鲜为人知的低秩(LR)恢复问题:从$n \times q$秩- $r$矩阵$X^* =[x^*_1 , x^*_2, ..., x^*_q]$的每个$q$列的$m$独立线性投影中恢复具有$r \ll \min(n,q)$的$n \times q$秩- $r$矩阵,即,当$y_k$是具有$m < n$的$m$长度向量时,从$y_k := A_k x^*_k , k \in [q]$。矩阵$A_k$是已知的,并且对于不同的$k$是相互独立的。我们介绍了一种新的梯度下降(GD)为基础的解决方案称为AltGD-Min。用i.i.d.高斯项,并且如果$X^*$的右奇异向量满足不相干假设,则$X^*$的$ε$ -精确恢复是可能的,具有顺序$(n+q) r^2 \log(1/ε)$总样本和顺序$ mq nr \log (1/ε)$时间。与现有工作相比,这是最快的解决方案。对于$ε< r^{1/4}$,它还具有最佳的样本复杂度。AltGD-Min的一个简单扩展也可证明解决LR相位恢复,这是上述问题的仅幅度推广。AltGD-Min将未知的$X$分解为$X = UB$,其中$U$和$B$分别是具有$r$列和行的矩阵。它在用于更新$U$的(预计的)GD步骤和用于更新$B$的最小化步骤之间交替。它的每次迭代与常规投影GD一样快,因为$B$上的最小化是逐列的。同时,我们可以证明它的指数误差衰减,这是我们无法为投影GD。最后,它也可以有效地联合,每个节点的通信成本只有$nr$,而不是$nq$。

英文摘要:

We study the following lesser-known low rank (LR) recovery problem: recover an $n \times q$ rank-$r$ matrix, $X^* =[x^*_1 , x^*_2, ..., x^*_q]$, with $r \ll \min(n,q)$, from $m$ independent linear projections of each of its $q$ columns, i.e., from $y_k := A_k x^*_k , k \in [q]$, when $y_k$ is an $m$-length vector with $m < n$. The matrices $A_k$ are known and mutually independent for different $k$. We introduce a novel gradient descent (GD) based solution called AltGD-Min. We show that, if the $A_k$s are i.i.d. with i.i.d. Gaussian entries, and if the right singular vectors of $X^*$ satisfy the incoherence assumption, then $ε$-accurate recovery of $X^*$ is possible with order $(n+q) r^2 \log(1/ε)$ total samples and order $ mq nr \log (1/ε)$ time. Compared with existing work, this is the fastest solution. For $ε< r^{1/4}$, it also has the best sample complexity. A simple extension of AltGD-Min also provably solves LR Phase Retrieval, which is a magnitude-only generalization of the above problem. AltGD-Min factorizes the unknown $X$ as $X = UB$ where $U$ and $B$ are matrices with $r$ columns and rows respectively. It alternates between a (projected) GD step for updating $U$, and a minimization step for updating $B$. Its each iteration is as fast as that of regular projected GD because the minimization over $B$ decouples column-wise. At the same time, we can prove exponential error decay for it, which we are unable to for projected GD. Finally, it can also be efficiently federated with a communication cost of only $nr$ per node, instead of $nq$ for projected GD.

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。

↑