Optimal principal component analysis in distributed and streaming models

@inproceedings{Boutsidis2015OptimalPC,
  title={Optimal principal component analysis in distributed and streaming models},
  author={Christos Boutsidis and David P. Woodruff and Peilin Zhong},
  booktitle={STOC 2016},
  year={2015}
}
  • Christos Boutsidis, David P. Woodruff, Peilin Zhong
  • Published in STOC 2015
  • Mathematics, Computer Science
  • This paper studies the Principal Component Analysis (PCA) problem in the distributed and streaming models of computation. Given a matrix <b>A</b> ∈ <b>R</b><sup>m×n</sup>, a rank parameter k<rank(<b>A</b>), and an accuracy parameter 0<ε<1, we want to output an m×k orthonormal matrix <b>U</b> for which ||<b>A</b>-<b>UU</b><sup>T</sup><b>A</b>||<sup>2</sup><sub>F</sub>≤(1+ε)||<b>A</b>-<b>A</b><sub>k</sub>||<sup>2</sup><sub>F</sub> where <b>A</b><sub>k</sub>∈<b>R</b><sup>m×n</sup> is the best rank… CONTINUE READING

    Create an AI-powered research feed to stay up to date with new papers like this posted to ArXiv

    Citations

    Publications citing this paper.
    SHOWING 1-10 OF 46 CITATIONS

    Near Optimal Frequent Directions for Sketching Dense and Sparse Matrices

    VIEW 5 EXCERPTS
    CITES METHODS & BACKGROUND
    HIGHLY INFLUENCED

    Efficient Matrix Sketching over Distributed Data

    VIEW 9 EXCERPTS
    CITES BACKGROUND & METHODS
    HIGHLY INFLUENCED

    Fast and Space-optimal Low-rank Factorization in the Streaming Model With Application in Differential Privacy

    VIEW 5 EXCERPTS
    CITES BACKGROUND
    HIGHLY INFLUENCED

    Tighter bound of Sketched Generalized Matrix Approximation

    VIEW 5 EXCERPTS
    CITES BACKGROUND & METHODS
    HIGHLY INFLUENCED

    Pass-Efficient Randomized Algorithms for Low-Rank Matrix Approximation Using Any Number of Views

    VIEW 6 EXCERPTS
    CITES BACKGROUND & METHODS
    HIGHLY INFLUENCED

    An Improved Gap-Dependency Analysis of the Noisy Power Method

    VIEW 4 EXCERPTS
    CITES BACKGROUND
    HIGHLY INFLUENCED

    Distributed Stochastic Algorithms for High-rate Streaming Principal Component Analysis

    VIEW 3 EXCERPTS
    CITES BACKGROUND & METHODS

    References

    Publications referenced by this paper.
    SHOWING 1-7 OF 7 REFERENCES

    R

    • J. Poulson, B. Marker
    • A. van de Geijn, J. R. Hammond, and N. A. Romero. Elemental: A new framework for distributed memory dense matrix computations. TOMS, 39(2)
    • 2013
    VIEW 5 EXCERPTS
    HIGHLY INFLUENTIAL

    Relative Errors for Deterministic Low-Rank Matrix Approximations

    VIEW 8 EXCERPTS
    HIGHLY INFLUENTIAL

    Tighter Low-rank Approximation via Sampling the Leveraged Element

    VIEW 5 EXCERPTS
    HIGHLY INFLUENTIAL

    Simple and deterministic matrix sketching

    VIEW 7 EXCERPTS
    HIGHLY INFLUENTIAL