随机截断振幅流求解随机二次方程大系统
Solving Large-scale Systems of Random Quadratic Equations via Stochastic Truncated Amplitude Flow
摘要:
一种新的方法称为\emph{stochastic truncated amplitude flow}(STAF)开发重建一个未知的$n$维实/复值信号$\bm{x}$从$m$“无相”二次方程的形式$ψ_i=|\langle\bm{a}_i,\bm{x}\rangle|$。这个问题,也被称为从仅幅度信息的相位恢复,通常是\emph{NP-hard}。采用基于幅度的非凸公式,STAF导致迭代求解器,包括两个阶段:s1)通过随机方差降低梯度算法的随机性促进初始化;和,s2)一系列的迭代改进的初始化使用随机截断梯度迭代。这两个阶段涉及一个单一的方程每次迭代,从而呈现STAF一个简单的,可扩展的,快速的方法,适合大规模的实现,这是有用的,当$n$是大的。当$\{\bm{a}_i\}_{i=1}^m$是独立的高斯,STAF可证明恢复正是任何$\bm{x}\in\mathbb{R}^n$指数快速的基础上的顺序$n$二次方程。STAF在有界支持的加性噪声的存在下也是鲁棒的。涉及真实高斯$\{\bm{a}_i\}$向量的模拟测试表明,STAF凭经验从大约仅测量$2.3n$的幅度精确地重建任何$\bm{x}\in\mathbb{R}^n$,优于最先进的方法,并缩小了与信息理论方程数$m=2n-1$的差距。使用合成数据和真实图像的广泛实验证实了STAF比现有替代品显着改善的性能。
英文摘要:
A novel approach termed \emph{stochastic truncated amplitude flow} (STAF) is developed to reconstruct an unknown $n$-dimensional real-/complex-valued signal $\bm{x}$ from $m$ `phaseless' quadratic equations of the form $ψ_i=|\langle\bm{a}_i,\bm{x}\rangle|$. This problem, also known as phase retrieval from magnitude-only information, is \emph{NP-hard} in general. Adopting an amplitude-based nonconvex formulation, STAF leads to an iterative solver comprising two stages: s1) Orthogonality-promoting initialization through a stochastic variance reduced gradient algorithm; and, s2) A series of iterative refinements of the initialization using stochastic truncated gradient iterations. Both stages involve a single equation per iteration, thus rendering STAF a simple, scalable, and fast approach amenable to large-scale implementations that is useful when $n$ is large. When $\{\bm{a}_i\}_{i=1}^m$ are independent Gaussian, STAF provably recovers exactly any $\bm{x}\in\mathbb{R}^n$ exponentially fast based on order of $n$ quadratic equations. STAF is also robust in the presence of additive noise of bounded support. Simulated tests involving real Gaussian $\{\bm{a}_i\}$ vectors demonstrate that STAF empirically reconstructs any $\bm{x}\in\mathbb{R}^n$ exactly from about $2.3n$ magnitude-only measurements, outperforming state-of-the-art approaches and narrowing the gap from the information-theoretic number of equations $m=2n-1$. Extensive experiments using synthetic data and real images corroborate markedly improved performance of STAF over existing alternatives.
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。