mlsanityData to 14 Oct 2026

Optimal spectral initializers impact on phase retrieval phase transitions -- an RDT view

arXiv 2506.18279 ↗First posted 23 Jun 2025stat.ML cs.IT cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

We analyze the relation between spectral initializers and theoretical limits of descending phase retrieval algorithms (dPR). In companion paper [104], for any sample complexity ratio, α, parametric manifold, mathcal PM(α), is recognized as a critically important structure that generically determines dPRs abilities to solve phase retrieval (PR). Moreover, overlap between the algorithmic solution and the true signal is positioned as a key mathcal PM's component. We here consider the so-called overlap optimal spectral initializers (OptSpins) as dPR's starting points and develop a generic Random duality theory (RDT) based program to statistically characterize them. In particular, we determine the functional structure of OptSpins and evaluate the starting overlaps that they provide for the dPRs. Since mathcal PM's so-called flat regions are highly susceptible to local jitteriness and as such are key obstacles on dPR's path towards PR's global optimum, a precise characterization of the starting overlap allows to determine if such regions can be successfully circumvented. Through the presented theoretical analysis we observe two key points in that regard: textbf(i) dPR's theoretical phase transition (critical α above which they solve PR) might be difficult to practically achieve as the mathcal PM's flat regions are large causing the associated OptSpins to fall exactly within them; and textbf(ii) Opting for so-called ``safer compression'' and slightly increasing α (by say 15%) shrinks flat regions and allows OptSpins to fall outside them and dPRs to ultimately solve PR. Numerical simulations are conducted as well and shown to be in an excellent agreement with theoretical predictions.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 40 of its 115 references that are in the map. First posted on arXiv 23 Jun 2025.

Builds on

40 of 115 references.

References · 40

  1. 2017Compressed Sensing using Generative Modelscited by 339
  2. 2014Phase Retrieval via Wirtinger Flow: Theory and Algorithmscited by 125
  3. 2017Optimal Errors and Phase Transitions in High-Dimensional Generalized Linear Modelscited by 116
  4. 2020Exactly Computing the Local Lipschitz Constant of ReLU Networkscited by 102
  5. 2016A Geometric Analysis of Phase Retrievalcited by 92
  6. 2018Phase Retrieval Under a Generative Priorcited by 73
  7. 2017Implicit Regularization in Nonconvex Statistical Estimation: Gradient Descent Converges Linearly for Phase Retrieval, Matrix Completion, and Blind Deconvolutioncited by 63
  8. 2013Phase Retrieval using Alternating Minimizationcited by 58
  9. 2015Solving Random Quadratic Systems of Equations Is Nearly as Easy as Solving Linear Systemscited by 50
  10. 2015Subdominant Dense Clusters Allow for Simple Learning and High Computational Performance in Neural Networks with Discrete Synapsescited by 39
  11. 2017Fundamental Limits of Weak Recovery with Applications to Phase Retrievalcited by 39
  12. 2020Phase retrieval in high dimensions: Statistical and computational phase transitionscited by 39
28 more
  1. 2019Inverting Deep Generative models, One layer at a timecited by 33
  2. 2016Solving Systems of Random Quadratic Equations via Truncated Amplitude Flowcited by 29
  3. 2013Phase Retrieval from Coded Diffraction Patternscited by 26
  4. 2017Phase Transitions of Spectral Initialization for High-Dimensional Nonconvex Estimationcited by 26
  5. 2016PhaseMax: Convex Phase Retrieval via Basis Pursuitcited by 25
  6. 2013Regularly random dualitycited by 24
  7. 2013Another look at the Gardner problemcited by 24
  8. 2013Limits of local algorithms over sparse random graphscited by 23
  9. 2015Optimal Rates of Convergence for Noisy Sparse Phase Retrieval via Thresholded Wirtinger Flowcited by 23
  10. 2015Local entropy as a measure for sampling solutions in Constraint Satisfaction Problemscited by 20
  11. 2018Optimization of the Sherrington-Kirkpatrick Hamiltoniancited by 20
  12. 2019Exact asymptotics for phase retrieval and compressed sensing with random generative priorscited by 20
  13. 2023Fully lifted random duality theorycited by 18
  14. 2016Phase retrieval with random Gaussian sensing vectors by alternating projectionscited by 15
  15. 2017Phase Retrieval via Randomized Kaczmarz: Theoretical Guaranteescited by 15
  16. 2017Structured signal recovery from quadratic measurements: Breaking sample complexity barriers via nonconvex optimizationcited by 13
  17. 2019Clustering of solutions in the symmetric binary perceptroncited by 12
  18. 2016Phase Retrieval Meets Statistical Learning Theory: A Flexible Convex Relaxationcited by 11
  19. 2023Injectivity of ReLU networks: perspectives from statistical physicscited by 10
  20. 2020Constant-Expansion Suffices for Compressed Sensing with Generative Priorscited by 9
  21. 2013Sparse Phase Retrieval: Uniqueness Guarantees and Recovery Algorithmscited by 8
  22. 2014Alternating Projection, Ptychographic Imaging and Phase Synchronizationcited by 5
  23. 2015PhaseLift is robust to a constant fraction of arbitrary errorscited by 4
  24. 2024Injectivity capacity of ReLU gatescited by 4
  25. 2016An Elementary Proof of Convex Phase Retrieval in the Natural Parameter Space via the Linear Program PhaseMaxcited by 3
  26. 2016Box constrained ℓ1 optimization in random linear systems -- asymptoticscited by 3
  27. 2019Fixed Point Analysis of Douglas-Rachford Splitting for Ptychography and Phase Retrievalcited by 3
  28. 2024Deep ReLU networks -- injectivity capacity upper boundscited by 3

Built on by

Not cited yet.