Nonconvex Optimization Meets Low-Rank Matrix Factorization: An Overview
Open accessThe DOI matches its Crossref record; 267 of the 279 references it cites were found in a registry or an index.
Abstract
Substantial progress has been made recently on developing provably accurate and efficient algorithms for low-rank matrix factorization via nonconvex optimization. While conventional wisdom often takes a dim view of nonconvex optimization algorithms due to their susceptibility to spurious local minima, simple iterative methods such as gradient descent have been remarkably successful in practice. The theoretical footings, however, had been largely lacking until recently. In this tutorial-style overview, we highlight the important role of statistical models in enabling efficient nonconvex optimization with performance guarantees. We review two contrasting approaches: (1) two-stage algorithms, which consist of a tailored initialization step followed by successive refinement; and (2) global landscape analysis and initialization-free algorithms. Several canonical matrix factorization problems are discussed, including but not limited to matrix sensing, phase retrieval, matrix completion, blind deconvolution, and robust principal component analysis. Special care is taken to illustrate the key technical insights underlying their analyses. This article serves as a testament that the integrated consideration of optimization and statistics leads to fruitful research findings.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Figures
![Fig. 1 An example of f ( · ) taken from [21]. Here, f ( x ) = x 2 if x ∈ [ -6 , 6] and f ( x ) = x 2 +1 . 5 | x | (cos( | x | -6) -1) if | x | > 6 . It satisfies RC ( μ, λ , ζ ) with some μ, λ > 0 and ζ = ∞ , but is clearly nonconvex.](https://figures.cochat.ai/paper-figures/efaa47591b5349acbef65a17306cdcb7/nonconvex-optimization-meets-low-rank-matrix-factorization-an-overview-figure-1.png)
![Fig. 2 Illustration of the function f ( x ) = 1 4 ‖ xx ⊤ -M ‖ 2 F , where x = [ x 1 , x 2 ] ⊤ and M =[ 1 -0 . 5 -0 . 5 1 ] .](https://figures.cochat.ai/paper-figures/a0e28a4860794cd1847da664546e9c3b/nonconvex-optimization-meets-low-rank-matrix-factorization-an-overview-figure-2.png)

![Fig. 4 Comparisons of 3 designs of the preprocessing function for estimating a 64 × 64 cameraman image from phaseless measurements under Poisson noise. (Red curve) the subset scheme in (119), where the parameter γ has been optimized for each fixed α ; (Green curve) the function T ∗ α ( · ) defined in [129]; (Blue curve) the uniformly optimal design T ∗ ( · ) in [130].](https://figures.cochat.ai/paper-figures/e8c537207c1141a888de05d5b42b9ec8/nonconvex-optimization-meets-low-rank-matrix-factorization-an-overview-figure-4.png)
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Fortunately, under the high-dimensional statistical setting, the local strong convexity and smoothness of the nonconvex loss functions can be exploited to tame the nonconvexity for various learning models, e.g., blind demixing [189], phase retrieval [254], and shallow neural networks [255].”
From Edge artificial intelligence for 6g: vision, enabling technologies, and applications (2022) · cites this paper as [254]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Matrix Factorization Techniques for Recommender Systems
Koren, Y., Bell, R., & Volinsky, C. (2009) · Computer · doi:10.1109/mc.2009.263 · Cited by 11,673 · cited 1×
“However, despite its wide use and remarkable performance in practice [15], [16], the foundational understanding of generic nonconvex optimization is far from mature.”
BackgroundA Stochastic Approximation Method
Robbins, H., & Monro, S. (1951) · The Annals of Mathematical Statistics · doi:10.1214/aoms/1177729586 · Cited by 9,643
Robust principal component analysis?
Candès, E. J., Li, X., Ma, Y., & Wright, J. (2011) · Journal of the ACM · doi:10.1145/1970392.1970395 · Cited by 6,921 · cited 2×
“Examples include low-rank matrix completion [1]–[3], phase retrieval [4], blind deconvolution and self-calibration [5], [6], robust principal component analysis [7], [8], synchronization and alignment [9]–[11], to name just a few.”
BackgroundA method for solving the convex programming problem with convergence rate O(1/k^2)
Nesterov, Y. E. (1983) · Доклады Академии наук · Cited by 6,318
Phase retrieval algorithms: a comparison
Fienup, J. R. (1982) · Applied Optics · doi:10.1364/ao.21.002758 · Cited by 5,596 · cited 3×
“This is exactly the Error Reduction (ER) algorithm proposed by Gerchberg and Saxton [43, 112] in the 1970s.”
BackgroundMethodExact Matrix Completion via Convex Optimization
Candès, E. J., & Recht, B. (2009) · Foundations of Computational Mathematics · doi:10.1007/s10208-009-9045-5 · Cited by 4,877
A practical algorithm for the determination of phase from image and diffraction plane pictures
Gerchberg, R. W. (1972) · Optik · Cited by 4,519
Halko, N., Martinsson, P. G., & Tropp, J. A. (2011) · SIAM Review · doi:10.1137/090771806 · Cited by 4,002
The restricted isometry property and its implications for compressed sensing
Candès, E. J. (2008) · Comptes Rendus. Mathématique · doi:10.1016/j.crma.2008.03.014 · Cited by 3,686 · cited 2×
“Lemma 3 ([43]): If an operator A satisfies the 2r-RIP with RIP constant δ2r < 1, then ∣ ∣ 〈 A(X),A(Y ) 〉 − 〈X,Y 〉 ∣ ∣ ≤ δ2r‖X‖F‖Y ‖F (23)”
BackgroundIntroductory Lectures on Convex Optimization: A Basic Course
Nesterov, Y. E. (2014) · Medical Entomology and Zoology · Cited by 3,557
Research works printed but not linked: 6
- Y. Chi, Y. M. Lu, and Y. Chen, "Nonconvex optimization meets low-rank matrix factorization: An overview," arXiv:1809.09573, 2018. · arXiv
- Q. Qu, Y. Zhang, Y. Eldar, and J. Wright, "Convolutional phase retrieval," in Proc. Neural Inf. Process. Syst., 2017, pp. 6086-6096.
- Z. Allen-Zhu, "Recent advances in stochastic convex and non-convex optimization," ICML Tut., 2017. [Online]. Available: http://people.csail.mit.edu/zeyuan/topics/icml-2017.htm
- M. Soltanolkotabi, "Algorithms and theory for clustering and nonconvex quadratic programming," Ph.D. dissertation, Stanford Univ., Stanford, U.K., 2014.
- J. Duchi, S. Shalev-Shwartz, Y. Singer, and T. Chandra, "Efficient pro- jections onto the ℓ 1 -ball for learning in high dimensions," in Proc. Int. Conf. Mach. Learn., 2008, pp. 272-279.
- T. Zhao, Z. Wang, and H. Liu, "Nonconvex low rank matrix factorization via inexact first order oracle," in Proc. Adv. Neural Inf. Process. Syst., 2015, pp. 559-567.
Cited by
Items 301–390 of 390
Proximal dogleg opportunistic majorization for nonconvex and nonsmooth optimization
Zhou, Y., & Dai, W. (2026) · Signal Processing · doi:10.1016/j.sigpro.2026.110849 · Cited by 2
Solving Quadratic Equations via Amplitude-based Nonconvex Optimization
Monardo, V., Li, Y., & Chi, Y. (2019) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.2019.8682357 · Cited by 2
Self-citationOn the Sensitivity of Spectral Initialization for Noisy Phase Retrieval
Monardo, V., & Chi, Y. (2019) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.2019.8682584 · Cited by 2
“The spectral method [1, 2] is a popular approach for estimating x, where it is estimated via the top eigenvector (up to scaling) of a carefully constructed data matrix that is a sum of rank-one matrices aiai , each weighted by the corresponding measurement yi, or a function T (yi) of it.”
MethodSelf-citationGlobal Optimality Guarantees for Nonconvex Unsupervised Video Segmentation
Anderson, B. G., & Sojoudi, S. (2019) · Allerton Conference on Communication, Control, and Computing · doi:10.1109/allerton.2019.8919697 · Cited by 2
“These nonconvex approaches often permit parallelization, lending themselves to lowered computational cost and scalability to larger problems [13].”
BackgroundCompletion of Structured Low-Rank Matrices via Iteratively Reweighted Least Squares
Kümmerle, C., & Verdun, C. M. (2019) · International Conference on Sampling Theory and Applications · doi:10.1109/sampta45681.2019.9030959 · Cited by 2
“One benefit of these formulations is that it is possible to prove that certain tractable algorithms solve the underlying problems [5]–[7] if an optimal or near-optimal amount of samples is provided to the algorithms, and if the provided samples are generic or random enough.”
BackgroundLearning Latent Features with Pairwise Penalties in Low-Rank Matrix Completion
Ji, K., Tan, J. R., Xu, J., & Chi, Y. (2020) · IEEE Sensor Array and Multichannel Signal Processing Workshop (SAM) · doi:10.1109/sam48682.2020.9104354 · Cited by 2
“A typical optimization problem follows the form [3], [4], [5]”
BackgroundSelf-citationMatrix completion based adaptive sampling for measuring network delay with online support
Wei, M., & Li, L. (2020) · KSII Transactions on Internet and Information Systems · doi:10.3837/tiis.2020.07.018 · Cited by 2
“Internet Distance Estimation Service (IDES) [11] employs Singular Value Decomposition (SVD) or Nonnegative Matrix Factorization (NMF) to recover a delay matrix in a landmark-based manner.”
MethodImproved Regret Bounds of Bilinear Bandits using Action Space Analysis
Jang, K., Jun, K.-S., Yun, S.-Y., & Kang, W. (2021) · KAIST Institutional Repository (KAIST) · Cited by 2
Discrete-Aware Matrix Completion via Proximal Gradient
Iimori, H., de Abreu, G. T. F., Taghizadeh, O., & Ishibashi, K. (2020) · Asilomar Conference on Signals, Systems and Computers (ACSSC) · doi:10.1109/ieeeconf51394.2020.9443404 · Cited by 2
“For more information, please refer to a recent comprehensive survey [2] on non-convex MC problems and solutions.”
BackgroundLandscape Correspondence of Empirical and Population Risks in the Eigendecomposition Problem
Li, S., Tang, G., & Wakin, M. B. (2022) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2022.3181333 · Cited by 2
A Composite Initialization Method for Phase Retrieval
Luo, Q., Lin, S., & Wang, H. (2021) · Symmetry · doi:10.3390/sym13112006 · Cited by 2
“, the quadratic sensing and the matrix reconstruction problems [21].”
BackgroundVariational Bayesian Inference for CP Tensor Completion with Subspace Information
Budzinskiy, S., & Замарашкин, Н. Л. (2023) · Lobachevskii Journal of Mathematics · doi:10.1134/s1995080223080103 · Cited by 2
Chorobura, F., Lupu, D., & Necoara, I. (2022) · IEEE Conference on Decision and Control (CDC) · doi:10.1109/cdc51059.2022.9992996 · Cited by 2
“…a penalized formulation of the orthogonal nonnegative matrix factorization problem, used e.g., in dimensionality reduction of big data [2], [5]: Let us define the following functions: where For optimization problem (17) one can notice that the functions f ( W, V ) and ψ ( W, V ) are…”
Tracking the Adjacency Spectral Embedding for Streaming Graphs
Larroca, F., Bermolen, P., Fiori, M., Marenco, B., & Mateos, G. (2022) · Asilomar Conference on Signals, Systems and Computers (ACSSC) · doi:10.1109/ieeeconf56349.2022.10051861 · Cited by 2
“In this context, this approach is sometimes called factorized GD [18], or Procrustes flow [19], which has been shown to converge with linear rate to the solution X̂t of (3) if the initial value X̂t[0] is close [18], [20].”
BackgroundFast Gradient Method for Low-Rank Matrix Estimation
Li, H., Peng, Z., Pan, C., & Zhao, D. (2023) · Journal of Scientific Computing · doi:10.1007/s10915-023-02266-7 · Cited by 2
“Qualitative convergence analysis of this algorithm and its variants have been extensively studied, see [9].”
BackgroundInfluentialExtract Spatial Distribution of a Specific Gas from Mixed Gas Data Measured by the LSPR Gas Sensor
Zheng, X., Matsuoka, M., Hayashi, K., & Tomiura, Y. (2023) · Proceedings of IEEE Sensors/Proceedings of IEEE Sensors ... · doi:10.1109/sensors56945.2023.10324923 · Cited by 2
“Matrix decomposition [1] is an unsupervised learning al-gorithm, which decomposes a matrix into the product of two matrices.”
Oh, W., Jayaraman, P., Tandon, P., Chaddha, U., Kovatch, P., Charney, A. W., Glicksberg, B. S., & Nadkarni, G. N. (2023) · Artificial Intelligence in Medicine · doi:10.1016/j.artmed.2023.102750 · Cited by 2
Applications of Stochastic Gradient Descent to Nonnegative Matrix Factorization
Slavin, M. (2019) · UWSpace (University of Waterloo) · Cited by 3
Nearly Minimal Over-Parametrization of Shallow Neural NetworksPreprint
Eftekhari, A., Song, C., & Cevher, V. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1910.03948 · Cited by 2
“More generally, the idea of initializing an algorithm close to a global minimizer has a long history in nonconvex optimization, particularly matrix factorization problems, see the recent survey [31] and the many references therein.”
BackgroundNayer, S. S. (2021) · Iowa State University Digital Repository (Iowa State University) · doi:10.31274/etd-20210609-131 · Cited by 2
Zhang, Q., He, L., Wang, Y., Deng, L.-J., & Liu, J. (2023) · SSRN Electronic Journal · doi:10.2139/ssrn.4619731 · Cited by 2
Asymptotic Escape of Spurious Critical Points on the Low-rank Matrix ManifoldPreprint
Hou, T. Y., Li, Z., & Zhang, Z. (2021) · CaltechAUTHORS (California Institute of Technology) · doi:10.48550/arxiv.2107.09207 · Cited by 2
“In statistical estimation, this could mean a tendency to promote incoherence and accelerate convergence [3][14].”
BackgroundA random matrix framework for large dimensional machine learning and neural networks
Liao, Z. (2019) · theses.fr (ABES) · doi:10.70675/a524b5e8zcec6z4ed6z9975zd205cf648ff1 · Cited by 4
Zheng, X., Matsuoka, M., Hayashi, K., & Tomiura, Y. (2024) · IEEJ Transactions on Sensors and Micromachines · doi:10.1541/ieejsmas.144.345 · Cited by 1
Fast and Provable Hankel Tensor Completion for Multi-Measurement Spectral Compressed Sensing
Li, J., Zhang, X., Wu, S., & Cui, W. (2025) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2025.3607195 · Cited by 1
“Fast and nonconvex gradient methods based on low-rank factorization [14], [26], [27], [28] have garnered significant interest in recent years.”
Scalable Hessian-free Proximal Conjugate Gradient Method for Nonconvex and Nonsmooth Optimization
Zhou, Y., & Dai, W. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11461951 · Cited by 1
“Extending to the low-rank optimization [7], the smooth term typically has the form q ( X ) = ∥A ( X ) − Y ∥ 2 where A ( · ) is a linear operator.”
Moens, R., Migas, L. G., Anderson, D. M., Messinger, J. D., Ovchinnikova, O. S., Caprioli, R. M., Curcio, C. A., Schey, K. L., Spraggins, J. M., & de Plas, R. V. (2025) · Analytical Chemistry · doi:10.1021/acs.analchem.4c06368 · Cited by 1
Robust sparse phase retrieval: statistical guarantee, optimality theory and convergent algorithm
Fan, J., Yan, A., Xiu, X., & Liu, W. (2026) · Inverse Problems · doi:10.1088/1361-6420/ae5e02 · Cited by 1
“See the overview [50] for more details.”
Matrix Completion with Graph Information: A Provable Nonconvex Optimization Approach
Wang, Y., Yang, Y., Wang, K., Gao, S., & Liao, X. (2026) · INFORMS journal on computing · doi:10.1287/ijoc.2025.1169 · Cited by 1
PDRs4All: XXI. Haute Couture : Spectral stitching of JWST MIRI-IFU cubes with matrix completion
Canin, A., Févotte, C., Nicolas Dobigeon, B., Putte, D. V. D., Onaka, T., & Berné, O. (2026) · Astronomy and Astrophysics · doi:10.1051/0004-6361/202558120 · Cited by 1
Fast and Global Optimal Nonconvex Matrix Factorization via Perturbed Alternating Proximal Point
Lu, S., Hong, M., & Wang, Z. (2019) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.2019.8682941 · Cited by 1
“If all critical points of a function f(X) are either global minima or strict saddle points, we say that f(X) has benign landscape [25], which is the main property interested in this paper.”
BackgroundBlind Super-Resolution via Projected Gradient Descent
Mao, S., & Chen, J. (2022) · IEEE International Symposium on Information Theory (ISIT) · doi:10.1109/isit50566.2022.9834503 · Cited by 1
Efficient Implementation of Stochastic Proximal Point Algorithm for Matrix and Tensor Completion
Bumin, A., & Huang, K. (2021) · European Signal Processing Conference (EUSIPCO) · doi:10.23919/eusipco54536.2021.9616032 · Cited by 1
Fast Quadratic Sensing Via Nonconvex Optimization
Liu, K., Wan, L., & Sun, L. (2022) · European Signal Processing Conference (EUSIPCO) · doi:10.23919/eusipco55093.2022.9909632 · Cited by 1
“This problem is also known as quadratic sensing [4] and the scalar measurements are given as yi = ∥ai X∥(2) + ei, i = 1, .”
BackgroundWei, M., Chen, D., Yu, W., & Yao, L. (2023) · International Conference on Information Science and Technology · doi:10.1109/icist59754.2023.10367176 · Cited by 1
“By nature, this issue is ill-posed and non-linear since a multitude of signals can result in identical intensity measurements, making its resolution particularly challenging [2].”
BackgroundDiscrete Aware Matrix Completion via Convexized $\ell_{0}$-Norm Approximation
Führling, N., Ando, K., de Abreu, G. T. F., David González, G., & Gonsa, O. (2024) · Asilomar Conference on Signals, Systems and Computers (ACSSC) · doi:10.1109/ieeeconf60004.2024.10943107 · Cited by 1
“As an example, an excellent survey on non-convex optimization methods applied to low-rank MC can be found in [23].”
BackgroundLow-rank mirror-prox methods for nonsmooth and low-rank matrix optimization problems
Garber, D., & Kaplan, A. (2025) · Mathematical Programming · doi:10.1007/s10107-025-02267-4 · Cited by 1
ALPCAHUS: Subspace Clustering for Heteroscedastic Data
Cavazos, J. S., Fessler, J. A., & Balzano, L. (2026) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2026.3729726 · Cited by 1
“To reduce computation and enforce a low-rank solution, the LR-ALPCAH variant [17] took inspiration from the matrix factorization literature [35] and factorized X ∈ R D × N ≈ LR where L ∈ R D × ˆ d and R ∈ R N × ˆ d for some rank estimate ˆ d .”
Diagonal Preconditioning: Theory and AlgorithmsPreprint
Qu, Z., Ye, Y., & Zhou, Z. (2020) · arXiv (Cornell University) · Cited by 1
The Nonconvex Geometry of Linear Inverse ProblemsPreprint
Eftekhari, A., & Esfahani, P. M. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2101.02776 · Cited by 1
“, the well-known BurerMonteiro factorization for certain matrix- or tensor-valued learning problems, see [19, 42, 20].”
BackgroundRobust Matrix Factorization with Grouping EffectPreprint
Jiang, H., Li, S., Zhang, L., Xiong, H., & Dou, D. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2106.13681 · Cited by 1
“Matrix factorization (MF) is a fundamental yet widely-used technique in machine learning (Chi et al., 2019, Choi, 2008, Dai et al., 2020, Li et al., 2019, Mnih and Salakhutdinov, 2007, Parvin et al., 2019, Srebro et al., 2005, Wang et al., 2020), with numerous applications in computer vision (Gao…”
MethodLow Rank Vectorized Hankel Lift for Matrix Recovery via Fast Iterative Hard Thresholding.Preprint
Zhu, Z., Chen, J., & Gao, W. (2021) · arXiv (Cornell University) · Cited by 1
2019 Index IEEE Transactions on Signal Processing Vol. 67
(2019) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2020.2968163 · Cited by 1
Jian, A., Li, X., Sun, W., & Yu, G. (2025) · doi:10.36227/techrxiv.174612953.34891083/v1 · Cited by 1
TReF-6: Inferring Task-Relevant Frames from a Single Demonstration for One-Shot Skill GeneralizationPreprint
Ding, Y., Wang, S., & Fitzgerald, T. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2509.00310 · Cited by 2
“Empirically, we find that choosing an initialization point with (1) sharp gradients and (2) proximity to the ground truth , is essential for successful optimization - an insight aligned with prior works in optimization [38, 39].”
Doubling the dimension yields a benign landscape for the squared-stressPreprint
Criscitiello, C. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2608.16799 · Cited by 1
“EDG is an instance of low-rank matrix sensing (Chi et al., 2019), with positive semidefinite measurement matrices ( e i − e j )( e i − e j ) ⊤ for { i, j } ∈ E .”
Hyperspectral Image Restoration for Non-additive Noise
Zhang, F., Chang, H., Yang, Z., Zhan, T., & Yang, G. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-60633-6_19
Fattahi, S. (2020) · eScholarship (California Digital Library)
An Experimental Study of Balance in Matrix Factorization
Hsia, J., & Ramadge, P. J. (2021) · Conference on Information Sciences and Systems · doi:10.1109/ciss50987.2021.9400232
Phase Retrieval in Acoustical Signal Processing
Yatabe, K. (2021) · IEICE ESS FUNDAMENTALS REVIEW · doi:10.1587/essfr.15.1_25
Successful Recovery Performance Guarantees of Noisy SOMPPreprint
Zhang, W., & Kim, T. (2021) · arXiv (Cornell University)
A Study on Weighted and Structured Low Rank Matrix Optimization Problems
Shen, J. (2021) · ePrints Soton (University of Southampton)
Enhancing Matrix Factorization-based Recommender Systems via Graph Neural Networks
Guo, Z., Meng, D., Zhang, H., Wang, H., & Yu, K. (2021) · IEEE International Symposium on Parallel and Distributed Processing with Applications (ISPA) · doi:10.1109/ispa-bdcloud-socialcom-sustaincom52081.2021.00146 · Cited by 2
Over-Parametrized Matrix Factorization in the Presence of Spurious Stationary Points
Eftekhari, A. (2021) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2021.3139213 · Cited by 1
“1) (UNDER-PARAMETRIZED REGIME) When p À m{d, the landscape of problem (1) is often benign in the sense that problem (1), even though nonconvex, has no spurious stationary points [6].”
BackgroundMethodInfluentialEscosio, R. A. S., & Mendoza, R. (2022) · SSRN Electronic Journal · doi:10.2139/ssrn.4069780
Shi, Y., Dong, J., & Zhang, J. (2020) · Springer eBooks · doi:10.1007/978-981-15-3870-4_3
Black-Box Acceleration of Monotone Convex Program Solvers
London, P., Vardi, S., Eghbali, R., & Wierman, A. (2022) · Operations Research · doi:10.1287/opre.2022.2352
Robust Low-Rank Matrix Completion via a New Sparsity-Inducing RegularizerPreprint
Wang, Z.-Y., So, H. C., & Zoubir, A. M. (2023) · doi:10.36227/techrxiv.24265078
Robust Low-Rank Matrix Completion via a New Sparsity-Inducing RegularizerPreprint
Wang, Z.-Y., So, H. C., & Zoubir, A. M. (2023) · doi:10.36227/techrxiv.24265078.v1
Nonconvex deterministic matrix completion by projected gradient descent methods
Xu, H., Li, S., & Lin, J. (2026) · Journal of Complexity · doi:10.1016/j.jco.2026.102056
Supervised Factor Modeling for High-Dimensional Linear Time SeriesPreprint
Huang, F., Lu, K., & Li, G. (2024) · SSRN Electronic Journal · doi:10.2139/ssrn.4758811
A Mechanistic Study of Transformers Training DynamicsPreprint
Odonnat, A., Bouaziz, W., & Cabannes, V. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2410.24050
“Tokens processed by transformers interact through the attention module, which allows us to study their evolution as a dynamic system of interacting particles controlled by an ordinary differential equation (Geshkovski et al., 2023b, 2025; Lu et al., 2019).”
Geometric Analysis of Non-Convex Optimization Landscapes for Robust M-Estimation of Location
Yang, H., Zhao, Z., & Sun, Y. (2024) · IEEE Information Theory Workshop (ITW) · doi:10.1109/itw61385.2024.10806948
Deep Unfolding Low-Rank Factorization Network for mmWave Massive MIMO Channel Estimation
Li, P., Teng, Y., Yu, Y., Zhao, Y., & Liu, A. (2025) · IEEE Wireless Communications and Networking Conference · doi:10.1109/wcnc61545.2025.10978513
“One approach to significantly reduce computational complexity is to represent the matrix as a product of two low-rank factors and optimize these factor matrices [6].”
Cluster-infused low-rank subspace learning for robust multi-label classification
Zhu, Z., Zhou, C., Sun, S., Ntaye, E., Shen, X., & Liu, Z. (2025) · Applied Intelligence · doi:10.1007/s10489-025-06837-z
“Matrix factorization techniques [ 59–62 ] generalize low-rank representation by uncovering latent patterns and structures in the data.”
A High-Dimensional Statistical Theory for Convex and Nonconvex Matrix SensingPreprint
Agterberg, J., & Vidal, R. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2506.20659
Optical Diffractive Neural Networks for Matrix Completion
Yue, Z., Li, H., Wang, Y., Liu, Y., Yu, A., Hu, M., Wang, K., Mao, Y., & Qi, J. (2025) · International Conference on Microwave and Millimeter Wave Technology (ICMMT) · doi:10.1109/icmmt65948.2025.11187738
“In this study, we propose a matrix completion method based on optical diffractive neural networks [4].”
Recovering Wasserstein Distance Matrices from Few Measurements
Rana, M., Tasissa, A., Cai, H., Gavriyelov, Y., & Hamm, K. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11460676
“Given the computational challenges of convex methods, non-convex algorithms have also been explored, and similar recovery guarantees have been established in this setting [11].”
Bayesian singular value decomposition procedure based on the horseshoe prior
Wang, J., Gong, Y., Lin, H., Zhang, R., & Zhao, W. (2025) · Journal of Applied Statistics · doi:10.1080/02664763.2025.2589992
“In addition, Sun and Luo [32] and Chi et al. [10] also proposed to use matrix decomposition method to study the matrix completion problem.”
The Role of Rank in Mismatched Low-Rank Symmetric Matrix Estimation
Niu, P., Liu, Y., Fu, T., Fan, J., Deng, C., & Huang, Z. (2025) · IEEE Information Theory Workshop (ITW) · doi:10.1109/itw62417.2025.11240472
Sample Complexity of Branch-length Estimation by Maximum LikelihoodPreprint
Clancy, D., Lyu, H., & Roch, S. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2507.22038
“Recent work in statistical estimation theory has highlighted the importance of analyzing the geometric structure of likelihood landscapes (see, e.g., (Ma et al., 2020; Chen & Chen, 2019; Chi et al., 2019)).”
Stöger, D. (2019) · mediaTUM (Technical University of Munich)
On the Complexity of Optimization: Curved Spaces and Benign Landscapes
Criscitiello, C. (2025) · Infoscience (Ecole Polytechnique Fédérale de Lausanne) · doi:10.5075/epfl-thesis-11173 · Cited by 3
Fast Non-convex Matrix Sensing with Optimal Sample Complexity
Cai, J. F., Wu, T., & Xia, R. (2025) · Rare & Special e-Zone (The Hong Kong University of Science and Technology)
An efficient robust PCA guided spatio-temporal network for small target detection in infrared videos
Zhu, H., & Feng, X. (2026) · Optics & Laser Technology · doi:10.1016/j.optlastec.2025.114531 · Cited by 1
Liu, L., Han, C., Guo, T., & Liu, H. (2026) · Expert Systems with Applications · doi:10.1016/j.eswa.2026.131213
Li, X., & Liu, J. (2026) · Multimedia Systems · doi:10.1007/s00530-025-02180-y
“Non-convex penalty functions have shown great potential in the fields of sparse learning and low-rank matrix recovery, and recent studies have shown that they have been widely used in the fields of image processing, computer vision and machine learning [27, 28].”
Chen, J., & Xiang, Z. (2026) · Digital Signal Processing · doi:10.1016/j.dsp.2026.106007
A computational tour on random matrices : imaging through complex media and optical computing
Dong, J. (2020) · theses.fr (ABES) · doi:10.70675/05b1b0eez1c32z4c82z8b68zf7a2b58fac77 · Cited by 1
Efficient iterative methods for clustering and matching problems on graphs
Braun, G. (2022) · theses.fr (ABES) · doi:10.70675/85e5be1cz2930z4d3ezac20zf0498e33fece
Algorithmic and theoretical aspects of sparse deep neural networks
Le, Q.-T. (2023) · theses.fr (ABES) · doi:10.70675/01b91cd8zefbaz442az9887zbfcec3e1132f · Cited by 1
Sketching for inverse problems in image processing
Shi, H. (2023) · theses.fr (ABES) · doi:10.70675/530a598fz3bc4z4240zbf8czfbff652161d3
Yang, Y., Li, Z., Li, T., Dong, Y., & Huang, Y. (2026) · International Journal of Data Science and Analytics · doi:10.1007/s41060-026-01124-7
A Low-Rank Perspective on Similarity Matrix Completion
Ma, C., Yu, R., Chen, X., Zhang, Y., & Lei, Z. (2026) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2026.3702548
“Thus, problem (7) can be reformulated as: Note that, though the MF technique [74], [75], [76], [77] suggests that the weighting parameter λ of the regularizer can be set to 0, which leads to a smooth function with simpler gradients.”
A parameter computation and optimization method in multi-directional exciting tumor treating fields
Zhao, Y., Yue, S., & Fu, K. (2026) · Computer Methods and Programs in Biomedicine · doi:10.1016/j.cmpb.2026.109545
Huang, L., Deng, W., Li, N., & Zhang, R. (2026) · Circuits Systems and Signal Processing · doi:10.1007/s00034-026-03695-z
Geometric analysis of variational quantum eigensolver
Qin, Z. (2026) · Journal of Physics A Mathematical and Theoretical · doi:10.1088/1751-8121/ae8b60 · Cited by 2
STOD: Sparse Tensor Train Optimization via Orthogonal Decomposition for High-Dimensional Learning
Li, X., & Luo, Z. (2026) · Mathematics · doi:10.3390/math14173084
Are In-Context Images Worth 10 Dimensions?Preprint
de Senneville, A., Bou, X., Anger, J., von Gioi, R. G., & Facciolo, G. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2609.37659
“Nonetheless, the convergence of this algorithm has been studied previously (Chi et al., 2019). et al., 2014), Caltech101 (Fei-Fei et al., 2004), SUN397 (Xiao et al., 2010), OxfordPets (Parkhi et al., 2012), StanfordCars (Krause et al., 2013), Flowers102 (Nilsback & Zisserman, 2008), Food101…”
Coded Alternating Least Squares for Distributed Low-Rank Matrix Factorization
Wang, S., Yan, Q., Wu, K., Zhang, J., Wang, J., & Song, L. (2026) · IEEE Journal on Selected Areas in Communications · doi:10.1109/jsac.2026.3740372
Metadata
- Authors
- Chi, Y., Lu, Y. M., & Chen, Y.
- Affiliations
- Carnegie Mellon University (US)
- Harvard University (US)
- Princeton University (US)
- Venue
- IEEE Transactions on Signal Processing
- Published
- 2019-10-15
- Volume
- 67
- Issue
- 20
- Pages
- 5239-5269
- ISSN
- 1053-587X (ISSN-L)
- 1941-0476
- OpenAlex
- W2969215180
- Cited by
- 360
- References
- 279
- Citations by year
- 2026: 20 · 2025: 33 · 2024: 36 · 2023: 50 · 2022: 40 · 2021: 80 · 2020: 65 · 2019: 32
- FWCI
- 37.03
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- IEEE Transactions on Signal Processing (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- Office of Naval Research (N00014-18-1-2142, N00014-19-1-2404)
- Army Research Office (W911NF-18-1-0303)
- Air Force Office of Scientific Research (FA9550-15-1-0205)
- National Science Foundation (CCF-1901199, CCF-1806154, ECCS-1818571, CCF-1718698, CCF-1910410, CCF-1907661, IIS-1900140)
- Harvard Dean's Competitive Fund for Promising Research
- AFOSR (YIP FA9550-19-1-0030)
- ARO (W911NF-18-1-0303)
- ONR (N00014-19-1-2120)
- Topics
- Physical Sciences › Computer Science › Signal Processing › Sparse and Compressive Sensing Techniques
- Physical Sciences › Computer Science › Artificial Intelligence › Stochastic Gradient Optimization Techniques
- Physical Sciences › Mathematics › Numerical Analysis › Advanced Optimization Algorithms Research
- Keywords
- non-convex optimization, low-rank factorization, statistical models, spurious local minima, two-stage algorithm, matrix sensing, phase retrieval, matrix completion, blind deconvolution, robust principal component analysis, performance guarantees
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.