中文

通过分层稀疏恢复实现双稀疏盲去卷积

信息论 2024-11-12 v3 数值分析 math.IT 数值分析

摘要

分层稀疏框架,特别是 HiHTP 算法,近来已成功应用于许多相关的通信工程问题,尤其在信号空间具有分层结构时。本文研究 HiHTP 算法求解双稀疏盲去卷积问题的适用性。此处的双稀疏盲去卷积设定是从 h(Qb)h*(Qb) 的知识中恢复 hhbb,其中 QQ 为某线性算子,且 bbhh 均假设为稀疏。该方法依赖于将问题提升为线性问题,进而通过\emph{分层稀疏框架}应用 HiHTP。随后,对于随机矩阵 QQ 的高斯抽取,理论上证明当 μslog(s)2log(μ)log(μn)+sσlog(n)\mu \succcurlyeq s\log(s)^2\log(\mu)\log(\mu n) + s\sigma \log(n) 时,以高概率可恢复 ss-稀疏的 hKμh \in \mathbb{K}^\muσ\sigma-稀疏的 bKnb \in \mathbb{K}^n

关键词

引用

@article{arxiv.2210.11993,
  title  = {Bisparse Blind Deconvolution through Hierarchical Sparse Recovery},
  author = {Axel Flinth and Ingo Roth and Gerhard Wunder},
  journal= {arXiv preprint arXiv:2210.11993},
  year   = {2024}
}

备注

V3: Completely rewritten introduction, and a few corrections in the proof section. V2: Completely revised version, entirely different proof, resulting in the recovery guarantee improved by a factor s