Robust machine learning by median-of-means : theory and practice

@article{Lecu2017RobustML,
  title={Robust machine learning by median-of-means : theory and practice},
  author={Guillaume Lecu{\'e} and M. Lerasle},
  journal={arXiv: Statistics Theory},
  year={2017}
}
We introduce new estimators for robust machine learning based on median-of-means (MOM) estimators of the mean of real valued random variables. These estimators achieve optimal rates of convergence under minimal assumptions on the dataset. The dataset may also have been corrupted by outliers on which no assumption is granted. We also analyze these new estimators with standard tools from robust statistics. In particular, we revisit the concept of breakdown point. We modify the original definition… Expand
67 Citations
How Robust is the Median-of-Means? Concentration Bounds in Presence of Outliers
  • 1
Asymptotic normality of robust risk minimizers
  • PDF
Robust k-means Clustering for Distributions with Two Moments
  • 2
  • PDF
Robust subgaussian estimation with VC-dimension
  • 2
  • Highly Influenced
  • PDF
All-In-One Robust Estimator of the Gaussian Mean
  • 5
  • PDF
MONK - Outlier-Robust Mean Embedding Estimation by Median-of-Means
  • 14
  • PDF
ROBUST K-MEANS CLUSTERING FOR DISTRIBUTIONS WITH TWO MOMENTS
  • LEXEY, ROSHNIN, Ikita, HIVOTOVSKIY
  • 2020
Adaptive Hard Thresholding for Near-optimal Consistent Robust Regression
  • 19
  • PDF
Robust classification via MOM minimization
  • 28
  • PDF
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 114 REFERENCES
ROBUST EMPIRICAL MEAN ESTIMATORS
  • 58
  • PDF
Robust Low-Rank Matrix Estimation
  • 20
  • PDF
Challenging the empirical mean and empirical variance: a deviation study
  • 243
  • PDF
Estimator selection with respect to Hellinger-type risks
  • 36
  • PDF
Learning from MOM's principles: Le Cam's approach
  • 20
  • Highly Influential
  • PDF
Model selection via testing: an alternative to (penalized) maximum likelihood estimators
  • 130
  • PDF
Alternatives to the Median Absolute Deviation
  • 1,537
  • PDF
...
1
2
3
4
5
...