Corpus ID: 44070829

High Dimensional Robust Sparse Regression

@inproceedings{Liu2020HighDR,
  title={High Dimensional Robust Sparse Regression},
  author={L. Liu and Yanyao Shen and Tianyang Li and C. Caramanis},
  booktitle={AISTATS},
  year={2020}
}
  • L. Liu, Yanyao Shen, +1 author C. Caramanis
  • Published in AISTATS 2020
  • Computer Science, Mathematics
  • We provide a novel -- and to the best of our knowledge, the first -- algorithm for high dimensional sparse regression with constant fraction of corruptions in explanatory and/or response variables. Our algorithm recovers the true sparse parameters with sub-linear sample complexity, in the presence of a constant fraction of arbitrary corruptions. Our main contribution is a robust variant of Iterative Hard Thresholding. Using this, we provide accurate estimators: when the covariance matrix in… CONTINUE READING
    27 Citations
    Recent Advances in Algorithmic High-Dimensional Robust Statistics
    • 53
    • PDF
    Robust compressed sensing using generative models
    • PDF
    Robust compressed sensing of generative models
    • 4
    • PDF

    References

    SHOWING 1-10 OF 56 REFERENCES
    Computationally Efficient Robust Sparse Estimation in High Dimensions
    • 78
    • Highly Influential
    • PDF
    Robust Sparse Regression under Adversarial Corruption
    • 94
    • PDF
    High Dimensional Robust Estimation of Sparse Models via Trimmed Hard Thresholding
    • 12
    • PDF
    Consistent Robust Regression
    • 55
    • PDF
    High Dimensional Robust $M$-Estimation: Arbitrary Corruption and Heavy Tails
    • 8
    • PDF
    On Iterative Hard Thresholding Methods for High-dimensional M-Estimation
    • 159
    • PDF
    Efficient Algorithms for Outlier-Robust Regression
    • 76
    • Highly Influential
    • PDF
    Robust Estimation via Robust Gradient Estimation
    • 106
    • Highly Influential
    • PDF
    Outlier-Robust PCA: The High-Dimensional Case
    • 99
    • PDF
    Being Robust (in High Dimensions) Can Be Practical
    • 124
    • Highly Influential
    • PDF