POLYNOMIAL APPROXIMATION VIA COMPRESSED SENSING OF HIGH-DIMENSIONAL FUNCTIONS ON LOWER SETS

被引:37
|
作者
Chkifa, Abdellah [1 ]
Dexter, Nick [2 ]
Hoang Tran [1 ]
Webster, Clayton G. [1 ,2 ]
机构
[1] Oak Ridge Natl Lab, Dept Computat & Appl Math, 1 Bethel Valley Rd,POB 2008, Oak Ridge, TN 37831 USA
[2] Univ Tennessee, Dept Math, Knoxville, TN 37996 USA
关键词
Compressed sensing; high-dimensional methods; polynomial approximation; convex optimization; downward closed (lower) sets; PARTIAL-DIFFERENTIAL-EQUATIONS; STOCHASTIC COLLOCATION METHOD; RECONSTRUCTION; INTERPOLATION; CONVERGENCE; UNCERTAINTY; EXPANSIONS; ALGORITHMS; RECOVERY;
D O I
10.1090/mcom/3272
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
This work proposes and analyzes a compressed sensing approach to polynomial approximation of complex-valued functions in high dimensions. In this context, the target function is often smooth and characterized by a rapidly decaying orthonormal expansion, whose most important terms are captured by a lower (or downward closed) set. Motivated by this fact, we present an innovative weighted l(1)-minimization procedure with a precise choice of weights for imposing the downward closed preference. Theoretical results reveal that our computational approaches possess a provably reduced sample complexity compared to existing compressed sensing techniques presented in the literature. In addition, the recovery of the corresponding best approximation using these methods is established through an improved bound for the restricted isometry property. Our analysis represents an extension of the approach for Hadamard matrices by J. Bourgain [An improved estimate in the restricted isometry problem, Lecture Notes in Math., vol. 216, Springer, 2014, pp. 65-70] to the general bounded orthonormal systems, quantifies the dependence of sample complexity on the successful recovery probability, and provides an estimate on the number of measurements with explicit constants. Numerical examples are provided to support the theoretical results and demonstrate the computational efficiency of the novel weighted l(1)-minimization strategy.
引用
收藏
页码:1415 / 1450
页数:36
相关论文
共 50 条
  • [41] Visualization of very large high-dimensional data sets as minimum spanning trees
    Daniel Probst
    Jean-Louis Reymond
    Journal of Cheminformatics, 12
  • [42] Adaptive and Oblivious Randomized Subspace Methods for High-Dimensional Optimization: Sharp Analysis and Lower Bounds
    Lacotte, Jonathan
    Pilanci, Mert
    IEEE TRANSACTIONS ON INFORMATION THEORY, 2022, 68 (05) : 3281 - 3303
  • [43] Gaussian Approximation and Spatially Dependent Wild Bootstrap for High-Dimensional Spatial Data
    Kurisu, Daisuke
    Kato, Kengo
    Shao, Xiaofeng
    JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 2024, 119 (547) : 1820 - 1832
  • [44] Clustering High-Dimensional Data via Random Sampling and Consensus
    Traganitis, Panagiotis A.
    Slavakis, Konstantinos
    Giannakis, Georgios B.
    2014 IEEE GLOBAL CONFERENCE ON SIGNAL AND INFORMATION PROCESSING (GLOBALSIP), 2014, : 307 - 311
  • [45] LIMITING BEHAVIOR OF EIGENVALUES IN HIGH-DIMENSIONAL MANOVA VIA RMT
    Bai, Zhidong
    Choi, Kwok Pui
    Fujikoshi, Yasunori
    ANNALS OF STATISTICS, 2018, 46 (06) : 2985 - 3013
  • [46] High-Dimensional Nonconvex Stochastic Optimization by Doubly Stochastic Successive Convex Approximation
    Mokhtari, Aryan
    Koppel, Alec
    IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2020, 68 : 6287 - 6302
  • [47] High-dimensional variable selection via low-dimensional adaptive learning
    Staerk, Christian
    Kateri, Maria
    Ntzoufras, Ioannis
    ELECTRONIC JOURNAL OF STATISTICS, 2021, 15 (01): : 830 - 879
  • [48] ESTIMATION OF SMOOTH FUNCTIONALS IN HIGH-DIMENSIONAL MODELS: BOOTSTRAP CHAINS AND GAUSSIAN APPROXIMATION
    Koltchinskii, Vladimir
    ANNALS OF STATISTICS, 2022, 50 (04) : 2386 - 2415
  • [49] Compressed Sensing of Underwater Acoustic Signals via Structured Approximation l0-Norm
    Wu, Fei-Yun
    Yang, Kunde
    Duan, Rui
    IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, 2018, 67 (09) : 8504 - 8513
  • [50] Dimension-wise integration of high-dimensional functions with applications to finance
    Griebel, Michael
    Holtz, Markus
    JOURNAL OF COMPLEXITY, 2010, 26 (05) : 455 - 489