Efficient sum-of-squares algorithms achieve information-theoretically optimal error for robust mean estimation for the full range of adversarial corruption rates below 1/2.
□ B Preliminaries B.1 Sum-of-Squares Proofs to Algorithms Let us denote by /u1D44Ba vector or matrix of /u1D45Bindeterminates over ℝ
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DS 1years
2024 1verdicts
ACCEPT 1representative citing papers
citing papers explorer
-
Outlier-robust Mean Estimation near the Breakdown Point via Sum-of-Squares
Efficient sum-of-squares algorithms achieve information-theoretically optimal error for robust mean estimation for the full range of adversarial corruption rates below 1/2.