arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1604.00845cs.DS
收藏

次线性时间内具有近最佳采样复杂度的任意常维稀疏傅里叶变换

Sparse Fourier Transform in Any Constant Dimension with Nearly-Optimal Sample Complexity in Sublinear Time

Michael Kapralov

摘要:

我们考虑的问题,计算$k$稀疏近似的傅立叶变换的长度$N$信号。我们的主要结果是一个随机算法计算这样的近似(即实现$\ell_2/\ell_2$稀疏恢复保证使用傅立叶测量)使用$O_d(k\log N\log\log N)$样本的信号在时域中运行的时间$O_d(k\log^{d+3} N)$,其中$d\geq 1$是傅立叶变换的维数。样本复杂度与非自适应算法的$Ω(k\log (N/k))$的下限匹配,这是由于对于任何$k\leq N^{1-δ}$的\cite{DIPW},对于恒定的$δ>0$,直到$O(\log\log N)$因子。在我们的工作之前,具有可比的样本复杂性$k\log N \log^{O(1)}\log N$和次线性运行时间的结果是已知的傅立叶变换线上的\cite{IKP},但对于任何维度$d\geq 2$以前已知的技术要么遭受多对数因子损失的样本复杂性或所需的$Ω(N)$运行时间。

英文摘要:

We consider the problem of computing a $k$-sparse approximation to the Fourier transform of a length $N$ signal. Our main result is a randomized algorithm for computing such an approximation (i.e. achieving the $\ell_2/\ell_2$ sparse recovery guarantees using Fourier measurements) using $O_d(k\log N\log\log N)$ samples of the signal in time domain that runs in time $O_d(k\log^{d+3} N)$, where $d\geq 1$ is the dimensionality of the Fourier transform. The sample complexity matches the lower bound of $Ω(k\log (N/k))$ for non-adaptive algorithms due to \cite{DIPW} for any $k\leq N^{1-δ}$ for a constant $δ>0$ up to an $O(\log\log N)$ factor. Prior to our work a result with comparable sample complexity $k\log N \log^{O(1)}\log N$ and sublinear runtime was known for the Fourier transform on the line \cite{IKP}, but for any dimension $d\geq 2$ previously known techniques either suffered from a polylogarithmic factor loss in sample complexity or required $Ω(N)$ runtime.

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。

↑