Matches in SemOpenAlex for { <https://semopenalex.org/work/W3200213585> ?p ?o ?g. }
- W3200213585 abstract "We study the fundamental problem of ReLU regression, where the goal is to fit Rectified Linear Units (ReLUs) to data. This supervised learning task is efficiently solvable in the realizable setting, but is known to be computationally hard with adversarial label noise. In this work, we focus on ReLU regression in the Massart noise model, a natural and well-studied semi-random noise model. In this model, the label of every point is generated according to a function in the class, but an adversary is allowed to change this value arbitrarily with some probability, which is {em at most} $eta < 1/2$. We develop an efficient algorithm that achieves exact parameter recovery in this model under mild anti-concentration assumptions on the underlying distribution. Such assumptions are necessary for exact recovery to be information-theoretically possible. We demonstrate that our algorithm significantly outperforms naive applications of $ell_1$ and $ell_2$ regression on both synthetic and real data." @default.
- W3200213585 created "2021-09-27" @default.
- W3200213585 creator A5027430783 @default.
- W3200213585 creator A5054966819 @default.
- W3200213585 creator A5090325538 @default.
- W3200213585 date "2021-09-10" @default.
- W3200213585 modified "2023-09-27" @default.
- W3200213585 title "ReLU Regression with Massart Noise" @default.
- W3200213585 cites W1976986203 @default.
- W3200213585 cites W2005482089 @default.
- W3200213585 cites W2012712694 @default.
- W3200213585 cites W2102714321 @default.
- W3200213585 cites W2110273389 @default.
- W3200213585 cites W2112507308 @default.
- W3200213585 cites W2119667497 @default.
- W3200213585 cites W2125908420 @default.
- W3200213585 cites W2129249398 @default.
- W3200213585 cites W2141042406 @default.
- W3200213585 cites W2475016247 @default.
- W3200213585 cites W2558384550 @default.
- W3200213585 cites W2751645773 @default.
- W3200213585 cites W2755098789 @default.
- W3200213585 cites W2897521549 @default.
- W3200213585 cites W2926417365 @default.
- W3200213585 cites W2962703558 @default.
- W3200213585 cites W2962933442 @default.
- W3200213585 cites W2963351358 @default.
- W3200213585 cites W2963569411 @default.
- W3200213585 cites W2963589833 @default.
- W3200213585 cites W2963827833 @default.
- W3200213585 cites W2963874210 @default.
- W3200213585 cites W2963888996 @default.
- W3200213585 cites W2964008913 @default.
- W3200213585 cites W2964013305 @default.
- W3200213585 cites W2964135521 @default.
- W3200213585 cites W2965465787 @default.
- W3200213585 cites W2970031466 @default.
- W3200213585 cites W2970584880 @default.
- W3200213585 cites W2975778934 @default.
- W3200213585 cites W3023160755 @default.
- W3200213585 cites W3046237351 @default.
- W3200213585 cites W3046385591 @default.
- W3200213585 cites W3046762691 @default.
- W3200213585 cites W3091852560 @default.
- W3200213585 cites W3099625594 @default.
- W3200213585 cites W3100344698 @default.
- W3200213585 cites W3102827242 @default.
- W3200213585 cites W3103187151 @default.
- W3200213585 cites W3103933080 @default.
- W3200213585 cites W3112170761 @default.
- W3200213585 cites W3127251975 @default.
- W3200213585 cites W3127624461 @default.
- W3200213585 cites W3167082118 @default.
- W3200213585 cites W3177410885 @default.
- W3200213585 cites W3177530435 @default.
- W3200213585 cites W3198459381 @default.
- W3200213585 cites W616376197 @default.
- W3200213585 hasPublicationYear "2021" @default.
- W3200213585 type Work @default.
- W3200213585 sameAs 3200213585 @default.
- W3200213585 citedByCount "0" @default.
- W3200213585 crossrefType "posted-content" @default.
- W3200213585 hasAuthorship W3200213585A5027430783 @default.
- W3200213585 hasAuthorship W3200213585A5054966819 @default.
- W3200213585 hasAuthorship W3200213585A5090325538 @default.
- W3200213585 hasConcept C105795698 @default.
- W3200213585 hasConcept C11413529 @default.
- W3200213585 hasConcept C115961682 @default.
- W3200213585 hasConcept C119857082 @default.
- W3200213585 hasConcept C120665830 @default.
- W3200213585 hasConcept C121332964 @default.
- W3200213585 hasConcept C14036430 @default.
- W3200213585 hasConcept C152877465 @default.
- W3200213585 hasConcept C154945302 @default.
- W3200213585 hasConcept C162324750 @default.
- W3200213585 hasConcept C187736073 @default.
- W3200213585 hasConcept C192209626 @default.
- W3200213585 hasConcept C2524010 @default.
- W3200213585 hasConcept C2777212361 @default.
- W3200213585 hasConcept C2780451532 @default.
- W3200213585 hasConcept C28719098 @default.
- W3200213585 hasConcept C33923547 @default.
- W3200213585 hasConcept C41008148 @default.
- W3200213585 hasConcept C48921125 @default.
- W3200213585 hasConcept C78458016 @default.
- W3200213585 hasConcept C83546350 @default.
- W3200213585 hasConcept C86803240 @default.
- W3200213585 hasConcept C99498987 @default.
- W3200213585 hasConceptScore W3200213585C105795698 @default.
- W3200213585 hasConceptScore W3200213585C11413529 @default.
- W3200213585 hasConceptScore W3200213585C115961682 @default.
- W3200213585 hasConceptScore W3200213585C119857082 @default.
- W3200213585 hasConceptScore W3200213585C120665830 @default.
- W3200213585 hasConceptScore W3200213585C121332964 @default.
- W3200213585 hasConceptScore W3200213585C14036430 @default.
- W3200213585 hasConceptScore W3200213585C152877465 @default.
- W3200213585 hasConceptScore W3200213585C154945302 @default.
- W3200213585 hasConceptScore W3200213585C162324750 @default.
- W3200213585 hasConceptScore W3200213585C187736073 @default.
- W3200213585 hasConceptScore W3200213585C192209626 @default.