Matches in SemOpenAlex for { <https://semopenalex.org/work/W2964013305> ?p ?o ?g. }
Showing items 1 to 89 of
89
with 100 items per page.
- W2964013305 endingPage "2754" @default.
- W2964013305 startingPage "2745" @default.
- W2964013305 abstract "We study the prototypical problem of high-dimensional linear regression in a robust model where an e-fraction of the samples can be adversarially corrupted. We focus on the fundamental setting where the covariates of the uncorrupted samples are drawn from a Gaussian distribution N(0, Σ) on Rd. We give nearly tight upper bounds and computational lower bounds for this problem. Specifically our main contributions are as follows:• For the case that the covariance matrix is known to be the identity we give a sample near-optimal and computationally efficient algorithm that draws O(d/e2) labeled examples and outputs a candidate hypothesis vector [MATH HERE] that approximates the unknown regression vector β within e2-norm O(e log(1/e)σ), where σ is the standard deviation of the random observation noise. An error of Ω(eσ) is information-theoretically necessary even with infinite sample size. Hence, the error guarantee of our algorithm is optimal, up to a logarithmic factor in 1/e. Prior work gave an algorithm for this problem with sample complexity [MATH HERE] whose error guarantee scales with the e2-norm of β.• For the case of unknown covariance Σ, we show that we can efficiently achieve the same error guarantee of O(e log(1/e)σ), as in the known covariance case, using an additional O(d2 / e2) unlabeled examples. On the other hand, an error of O(eσ) can be information-theoretically attained with O(d/e2) samples. We prove a Statistical Query (SQ) lower bound providing evidence that this quadratic tradeoff in the sample size is inherent. More specifically, we show that any polynomial time SQ learning algorithm for robust linear regression (in Huber's contamination model) with estimation complexity O(d2−c), where c > 0 is an arbitrarily small constant, must incur an error of [MATH HERE]." @default.
- W2964013305 created "2019-07-30" @default.
- W2964013305 creator A5054966819 @default.
- W2964013305 creator A5083699810 @default.
- W2964013305 creator A5091325988 @default.
- W2964013305 date "2019-01-06" @default.
- W2964013305 modified "2023-09-26" @default.
- W2964013305 title "Efficient algorithms and lower bounds for robust linear regression" @default.
- W2964013305 doi "https://doi.org/10.5555/3310435.3310605" @default.
- W2964013305 hasPublicationYear "2019" @default.
- W2964013305 type Work @default.
- W2964013305 sameAs 2964013305 @default.
- W2964013305 citedByCount "53" @default.
- W2964013305 countsByYear W29640133052018 @default.
- W2964013305 countsByYear W29640133052019 @default.
- W2964013305 countsByYear W29640133052020 @default.
- W2964013305 countsByYear W29640133052021 @default.
- W2964013305 countsByYear W29640133052022 @default.
- W2964013305 crossrefType "proceedings-article" @default.
- W2964013305 hasAuthorship W2964013305A5054966819 @default.
- W2964013305 hasAuthorship W2964013305A5083699810 @default.
- W2964013305 hasAuthorship W2964013305A5091325988 @default.
- W2964013305 hasConcept C105795698 @default.
- W2964013305 hasConcept C11413529 @default.
- W2964013305 hasConcept C121332964 @default.
- W2964013305 hasConcept C129844170 @default.
- W2964013305 hasConcept C129848803 @default.
- W2964013305 hasConcept C134306372 @default.
- W2964013305 hasConcept C163716315 @default.
- W2964013305 hasConcept C17744445 @default.
- W2964013305 hasConcept C178650346 @default.
- W2964013305 hasConcept C185142706 @default.
- W2964013305 hasConcept C191795146 @default.
- W2964013305 hasConcept C199539241 @default.
- W2964013305 hasConcept C2524010 @default.
- W2964013305 hasConcept C28826006 @default.
- W2964013305 hasConcept C33923547 @default.
- W2964013305 hasConcept C39927690 @default.
- W2964013305 hasConcept C48921125 @default.
- W2964013305 hasConcept C62520636 @default.
- W2964013305 hasConcept C77553402 @default.
- W2964013305 hasConceptScore W2964013305C105795698 @default.
- W2964013305 hasConceptScore W2964013305C11413529 @default.
- W2964013305 hasConceptScore W2964013305C121332964 @default.
- W2964013305 hasConceptScore W2964013305C129844170 @default.
- W2964013305 hasConceptScore W2964013305C129848803 @default.
- W2964013305 hasConceptScore W2964013305C134306372 @default.
- W2964013305 hasConceptScore W2964013305C163716315 @default.
- W2964013305 hasConceptScore W2964013305C17744445 @default.
- W2964013305 hasConceptScore W2964013305C178650346 @default.
- W2964013305 hasConceptScore W2964013305C185142706 @default.
- W2964013305 hasConceptScore W2964013305C191795146 @default.
- W2964013305 hasConceptScore W2964013305C199539241 @default.
- W2964013305 hasConceptScore W2964013305C2524010 @default.
- W2964013305 hasConceptScore W2964013305C28826006 @default.
- W2964013305 hasConceptScore W2964013305C33923547 @default.
- W2964013305 hasConceptScore W2964013305C39927690 @default.
- W2964013305 hasConceptScore W2964013305C48921125 @default.
- W2964013305 hasConceptScore W2964013305C62520636 @default.
- W2964013305 hasConceptScore W2964013305C77553402 @default.
- W2964013305 hasLocation W29640133051 @default.
- W2964013305 hasOpenAccess W2964013305 @default.
- W2964013305 hasPrimaryLocation W29640133051 @default.
- W2964013305 hasRelatedWork W2046033161 @default.
- W2964013305 hasRelatedWork W2554864439 @default.
- W2964013305 hasRelatedWork W2597655115 @default.
- W2964013305 hasRelatedWork W2739524504 @default.
- W2964013305 hasRelatedWork W2751645773 @default.
- W2964013305 hasRelatedWork W2808754399 @default.
- W2964013305 hasRelatedWork W2901071012 @default.
- W2964013305 hasRelatedWork W2962820675 @default.
- W2964013305 hasRelatedWork W2962856918 @default.
- W2964013305 hasRelatedWork W2963351358 @default.
- W2964013305 hasRelatedWork W2963589833 @default.
- W2964013305 hasRelatedWork W2963945243 @default.
- W2964013305 hasRelatedWork W2964008913 @default.
- W2964013305 hasRelatedWork W2964135521 @default.
- W2964013305 hasRelatedWork W2964164735 @default.
- W2964013305 hasRelatedWork W2964288084 @default.
- W2964013305 hasRelatedWork W2970080249 @default.
- W2964013305 hasRelatedWork W2984272935 @default.
- W2964013305 hasRelatedWork W3017335724 @default.
- W2964013305 hasRelatedWork W616376197 @default.
- W2964013305 isParatext "false" @default.
- W2964013305 isRetracted "false" @default.
- W2964013305 magId "2964013305" @default.
- W2964013305 workType "article" @default.