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 1902.00131cs.ITmath.IT
收藏

高性能quantization可实现光谱超分辨率

High-performance quantization for spectral super-resolution

C. Sinan Güntürk, Weilin Li

摘要:

我们表明,分布式噪声整形β量化的方法提供了优越的性能的频谱超分辨率与量化的问题时,有冗余的测量数量。更确切地说,如果(整数)过采样比$λ$使得$\lfloor M/λ\rfloor - 1\geq 4/Δ$,其中$M$表示傅立叶测量的数量,$Δ$是与要解析的原子测量相关联的最小分离距离,则对于可用于测量的实部和虚部的任何数量的量化级$K\geq 2$,我们的量化方法保证了$O(λK^{- λ/2})$阶的重构精度,直到独立于$K$和$λ$的常数。相比之下,无记忆标量量化仅提供阶数$O(K^{-1})$的保证。

英文摘要:

We show that the method of distributed noise-shaping beta-quantization offers superior performance for the problem of spectral super-resolution with quantization whenever there is redundancy in the number of measurements. More precisely, if the (integer) oversampling ratio $λ$ is such that $\lfloor M/λ\rfloor - 1\geq 4/Δ$, where $M$ denotes the number of Fourier measurements and $Δ$ is the minimum separation distance associated with the atomic measure to be resolved, then for any number $K\geq 2$ of quantization levels available for the real and imaginary parts of the measurements, our quantization method guarantees reconstruction accuracy of order $O(λK^{- λ/2})$, up to constants which are independent of $K$ and $λ$. In contrast, memoryless scalar quantization offers a guarantee of order $O(K^{-1})$ only.

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。

↑