Solving Systems of Random Quadratic Equations via Truncated Amplitude Flow

@article{Wang2018SolvingSO,
  title={Solving Systems of Random Quadratic Equations via Truncated Amplitude Flow},
  author={G. Wang and Georgios B. Giannakis and Yonina C. Eldar},
  journal={IEEE Transactions on Information Theory},
  year={2018},
  volume={64},
  pages={773-794}
}
This paper presents a new algorithm, termed <italic>truncated amplitude flow</italic> (TAF), to recover an unknown vector <inline-formula> <tex-math notation="LaTeX">$ {x}$ </tex-math></inline-formula> from a system of quadratic equations of the form <inline-formula> <tex-math notation="LaTeX">$y_{i}=|\langle {a}_{i}, {x}\rangle |^{2}$ </tex-math></inline-formula>, where <inline-formula> <tex-math notation="LaTeX">$ {a}_{i}$ </tex-math></inline-formula>’s are given random measurement vectors… 
Scalable Solvers of Random Quadratic Equations via Stochastic Truncated Amplitude Flow
TLDR
STAF is a simple, scalable, and fast approach amenable to large-scale implementations that are useful to magnitude-only measurements, outperforming state-of-the-art approaches and narrowing the gap from the information-theoretic number of equations.
Solving Complex Quadratic Systems With Full-Rank Random Matrices
TLDR
This paper proves that the general class of problems with rotation-invariant sub-Gaussian measurement models can be efficiently solved with high probability via the standard framework comprising a spectral initialization followed by iterative Wirtinger flow updates on a nonconvex loss.
Perturbed Amplitude Flow for Phase Retrieval
TLDR
Perturbed Amplitude Flow algorithm solves a new proposed model, perturbed amplitude-based model, for phase retrieval, and is correspondingly named as PAF, and it is proved that PAF can recover LaTeX signals from phaseless samples.
Phase Retrieval via Reweighted Amplitude Flow
TLDR
The novel scheme is shown to be able to recover the true solution of the generalized phase retrieval problem in time proportional to reading the data in the high-dimensional regime given only an information-theoretic limit number of noiseless equations.
Structured Signal Recovery From Quadratic Measurements: Breaking Sample Complexity Barriers via Nonconvex Optimization
TLDR
The projected gradient descent, when initialized in a neighborhood of the desired signal, converges to the unknown signal at a linear rate and is proved to be the first provably tractable algorithm for this data-poor regime.
Toward the Optimal Construction of a Loss Function Without Spurious Local Minima for Solving Quadratic Equations
TLDR
Under the Gaussian measurement model, it is established that with high probability the target solution is the only minimizer of the new loss function provided by the smooth quadratic loss function with an activation function.
The Estimation Performance of Nonlinear Least Squares for Phase Retrieval
TLDR
The estimation performance of the model is presented with proving that the first theoretical guarantee for the nonlinear least squares and for thenonlinear Lasso of phase retrieval is presented.
Single-Index Models in the High Signal Regime
TLDR
A computationally efficient, end-to-end algorithm that achieves very fast rates of parameter estimation in the regime in which the signal- to-noise ratio in the problem is large is shown, significantly improves upon the risk of classical semi-parametric procedures.
Complete Dictionary Recovery Over the Sphere II: Recovery by Riemannian Trust-Region Method
TLDR
A Riemannian trust region algorithm that provably converges to a local minimizer with from arbitrary initializations is described and showed that with high probability, the nonconvex formulation has no “spurious” local minimizers and around any saddle point, the objective function has a negative directional curvature.
Optimization-Based AMP for Phase Retrieval: The Impact of Initialization and $\ell_{2}$ Regularization
TLDR
The sharp analyses in this paper enable us to compare the performance of the method with other phase recovery schemes but also shed light on designing better iterative algorithms for other non-convex optimization problems.
...
...

References

SHOWING 1-10 OF 100 REFERENCES
Scalable Solvers of Random Quadratic Equations via Stochastic Truncated Amplitude Flow
TLDR
STAF is a simple, scalable, and fast approach amenable to large-scale implementations that are useful to magnitude-only measurements, outperforming state-of-the-art approaches and narrowing the gap from the information-theoretic number of equations.
Solving Random Systems of Quadratic Equations via Truncated Generalized Gradient Flow
TLDR
It is proved that as soon as the number of equations $m$ is on the order of the numberof unknowns $n, TGGF recovers the solution exactly (up to a global unimodular constant) with high probability and complexity growing linearly with the time required to read the data.
Solving Most Systems of Random Quadratic Equations
TLDR
A novel procedure, that starts with a weighted maximal correlation initialization obtainable with a few power iterations, followed by successive refinements based oniteratively reweighted gradient-type iterations, which returns perfect signal recovery in the high-dimensional regime given only an information-theoretic limit number of equations.
The Local Convexity of Solving Systems of Quadratic Equations
TLDR
This paper shows that once the authors have m≥Cnrlog2(n) samples from isotropic gaussian samples, this implies that gradient descent with initialization (but no re-sampling) will converge linearly to the correct X, up to an orthogonal transformation.
Solving Almost all Systems of Random Quadratic Equations
TLDR
A novel procedure for minimizing the amplitude-based least-squares empirical loss is put forth, that starts with a weighted maximal correlation initialization obtainable with a few power or Lanczos iterations, followed by successive refinements based upon a sequence of iteratively reweighted (generalized) gradient iterations.
Sparse Phase Retrieval: Uniqueness Guarantees and Recovery Algorithms
TLDR
It is shown that TSPR can provably recover most discrete-time sparse signals and the recovery is robust in the presence of measurement noise, and these recovery guarantees are asymptotic in nature.
A Geometric Analysis of Phase Retrieval
TLDR
It is proved that when the measurement vectors are generic, with high probability, a natural least-squares formulation for GPR has the following benign geometric structure: (1) There are no spurious local minimizers, and all global minimizers are equal to the target signal, up to a global phase, and (2) the objective function has a negative directional curvature around each saddle point.
Solving Quadratic Equations via PhaseLift When There Are About as Many Equations as Unknowns
TLDR
It is shown that any complex vector can be recovered exactly from on the order of n quadratic equations of the form |〈ai,x0〉|2=bi, i=1,…,m, by using a semidefinite program known as PhaseLift, improving upon earlier bounds.
Phase recovery, MaxCut and complex semidefinite programming
TLDR
This work casts the phase retrieval problem as a non-convex quadratic program over a complex phase vector and formulates a tractable relaxation similar to the classical MaxCut semidefinite program.
PhaseLift: Exact and Stable Signal Recovery from Magnitude Measurements via Convex Programming
TLDR
It is shown that in some instances, the combinatorial phase retrieval problem can be solved by convex programming techniques, and it is proved that the methodology is robust vis‐à‐vis additive noise.
...
...