Publication (by topic)

Neural collapse

• Beyond unconstrained features: neural collapse for shallow neural networks with general data.
  W. Hong and S. Ling, submitted, 2024. (arXiv version).

• Cross entropy versus label smoothing: a neural collapse perspective.
  L. Guo, K. Ross, Z. Zhao, A. George, S. Ling, Y. Xu, Z. Dong, submitted, 2024. (arXiv version).

• Neural collapse for unconstrained feature model under cross-entropy loss with imbalanced data.
  W. Hong, S. Ling, Journal of Machine Learning Research 25(192):1-48, 2024. (arXiv version)(Journal)(Talk recording on Youtube).

Optimization landscape in synchronization

• Local geometry determines global landscape in low-rank factorization for synchronization.
  S. Ling, submitted, 2023. (arXiv version)(Talking recording by IMA at UMN, on Youtube).

• Solving orthogonal group synchronization via convex and low-rank optimization: tightness and landscape analysis.
  S. Ling, Mathematical Programming, Series A, 200, 589–628, 2023. (arXiv version)(Final)

• On the critical coupling of the finite Kuramoto model on dense networks.
  S. Ling, submitted, 2020. (arXiv version)

• On the landscape of synchronization networks: a perspective from nonconvex optimization.
  S. Ling, R. Xu, A. S. Bandeira, SIAM Journal on Optimization, 29(3):1879-1907, 2019. (arXiv version)(Final)(Talk Recording at the CMO)
  

Nonconvex optimization in Procrustes problem and synchronization

• Generalized orthogonal Procrustes problem under arbitrary adversaries.
  S. Ling, submitted, 2024. (arXiv version) (significantly revised in 2024).

• Improved performance guarantees for orthogonal group synchronization via generalized power method.
  S. Ling, SIAM Journal on Optimization, 32(2):1018-1048, 2022. (arXiv version)(Final)

• Near-optimal bounds for generalized orthogonal Procrustes problem via generalized power method.
  S. Ling, Applied and Computational Harmonic Analysis, 66, 62-100, 2023. (arXiv version)(Final)(Talk Recording on Youtube)(Code demo)

High-dimensional inference

• Uncertainty quantification of spectral estimator and MLE for orthogonal group synchronization.
  Z. S. Zhong and S. Ling, submitted, 2024. (arXiv version).

Exactness of convex relaxation in data science

• On the exactness of SDP relaxation for quadratic assignment problem,
  S. Ling, submitted, 2024. (arXiv version).

• Certifying global optimality of graph cuts via semidefinite relaxation: a performance guarantee for spectral clustering.
  S. Ling, T. Strohmer, Foundation of Computational Mathematics, 20(3):368-421, 2020. (arXiv version)(Final)(Slides)

• When do birds of a feather flock together? k-means, proximity, and conic programming.
  X. Li, Y. Li, S. Ling, T. Strohmer, K. Wei, Mathematical Programming, Series A, 179(1):295-341, 2020. (arXiv version)(Final)(Slides)

Spectral methods in data science

• Improved theoretical guarantee for rank aggregation via spectral method.
  Z. S. Zhong, S. Ling, Information and Inference: A Journal of the IMA, 13(3):1-36, 2024. (arXiv version)(Final).

• Near-optimal performance bounds for orthogonal and permutation group synchronization via spectral methods.
  S. Ling, Applied and Computational Harmonic Analysis 60, 20-52, 2022. (arXiv version)(Final)

• Strong consistency, graph Laplacians, and the stochastic block model.
  S. Deng, S. Ling, T. Strohmer, Journal of Machine Learning Research, 22(117):1−44, 2021. (arXiv version)(Final)

Nonconvex optimization and mathematics of signal processing

• Regularized gradient descent: a nonconvex recipe for fast joint blind deconvolution and demixing.
  S. Ling, T. Strohmer, Information and Inference: A Journal of the IMA, 8(1):1-49, 2019. (arXiv version)(Final)(Slides)

• Rapid, robust, and reliable blind deconvolution via nonconvex optimization.
  X. Li, S. Ling, T. Strohmer, K. Wei, Applied and Computational Harmonic Analysis, (47)3:893-934, 2019. (arXiv version)(Final)(Slides)(Talk Recording at the CMO)

• Fast blind deconvolution and blind demixing via nonconvex optimization.
  S.Ling, T.Strohmer, International Conference on Sampling Theory and Applications (SampTA), pp.114-118, 2017. (Final)

• You can have it all – Fast algorithms for blind deconvolution, self-calibration, and demixing.
  S.Ling, T.Strohmer, Mathematics in Imaging, MW1C.1, 2017. (Final)

Convex optimization in mathematics of signal processing

• Learning from their mistakes: self-calibrating sensors.
  B. Friedlander, S. Ling, T. Strohmer, SIAM News, 52(2), 2019. (Final)

• Self-calibration and bilinear inverse problems via linear least squares.
  S. Ling, T. Strohmer, SIAM Journal on Imaging Sciences, 11(1):252-292, 2018. (arXiv)(Final)

• Blind deconvolution meets blind demixing: algorithms and performance bounds.
  S. Ling, T. Strohmer, IEEE Transactions on Information Theory, 63(7):4497-4520, July 2017. (arXiv version)(Final)(Slides)

• Simultaneous blind deconvolution and blind demixing via convex programming.
  S.Ling, T.Strohmer, 50th Asilomar Conference on Signals, Systems and Computers, pp.1223-1227, 2016. (Final)

• Self-calibration and biconvex compressive sensing.
  S. Ling, T. Strohmer, Inverse Problems, (31)11:115002, 2015. (arXiv version)(Final)(Slides)
  (SIAM Student Paper Award 2017)

Numerical linear algebra

• Backward error and perturbation bounds for high order Sylvester tensor equation.
  X. Shi, Y. Wei, S. Ling, Linear and Multilinear Algebra, 61(10):1436-1446, 2013. (Final)

Applications in biology

• A metric and its derived protein network for evaluation of ortholog database inconsistency.
  W. Yang, J. Ji, S. Ling, G. Fang, Submitted, 2022. (bioRxiv version)

Dissertation

• Bilinear Inverse Problems: Theory, Algorithms, and Applications.
  S.Ling, University of California Davis, 2017, (Manucript)(Slides)