REVIEW 1 cited by
Robust polynomial regression up to the information theoretic limit
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
abstract
We consider the problem of robust polynomial regression, where one receives samples $(x_i, y_i)$ that are usually within $\sigma$ of a polynomial $y = p(x)$, but have a $\rho$ chance of being arbitrary adversarial outliers. Previously, it was known how to efficiently estimate $p$ only when $\rho < \frac{1}{\log d}$. We give an algorithm that works for the entire feasible range of $\rho < 1/2$, while simultaneously improving other parameters of the problem. We complement our algorithm, which gives a factor 2 approximation, with impossibility results that show, for example, that a $1.09$ approximation is impossible even with infinitely many samples.
Forward citations
Cited by 1 Pith paper
-
Sparse Polynomial Regression under Anomalous Data
A new algorithm, TS-CRR, solves anomaly-filtered sparse polynomial regression through a MILP-to-QCQP-to-fractional-program reformulation with conic relaxation, claiming better computational properties and good benchma...
Discussion (0). Continue with ORCID to comment.