FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3213309760> ?p ?o ?g. }

• W3213309760 abstract "We study the problem of PAC learning halfspaces with Massart noise. Given labeled samples $(x, y)$ from a distribution $D$ on $mathbb{R}^{d} times { pm 1}$ such that the marginal $D_x$ on the examples is arbitrary and the label $y$ of example $x$ is generated from the target halfspace corrupted by a Massart adversary with flipping probability $eta(x) leq eta leq 1/2$, the goal is to compute a hypothesis with small misclassification error. The best known $mathrm{poly}(d, 1/epsilon)$-time algorithms for this problem achieve error of $eta+epsilon$, which can be far from the optimal bound of $mathrm{OPT}+epsilon$, where $mathrm{OPT} = mathbf{E}_{x sim D_x} [eta(x)]$. While it is known that achieving $mathrm{OPT}+o(1)$ error requires super-polynomial time in the Statistical Query model, a large gap remains between known upper and lower bounds. In this work, we essentially characterize the efficient learnability of Massart halfspaces in the Statistical Query (SQ) model. Specifically, we show that no efficient SQ algorithm for learning Massart halfspaces on $mathbb{R}^d$ can achieve error better than $Omega(eta)$, even if $mathrm{OPT} = 2^{-log^{c} (d)}$, for any universal constant $c in (0, 1)$. Furthermore, when the noise upper bound $eta$ is close to $1/2$, our error lower bound becomes $eta - o_{eta}(1)$, where the $o_{eta}(1)$ term goes to $0$ when $eta$ approaches $1/2$. Our results provide strong evidence that known learning algorithms for Massart halfspaces are nearly best possible, thereby resolving a longstanding open problem in learning theory." @default.
• W3213309760 created "2021-11-22" @default.
• W3213309760 creator A5035962853 @default.
• W3213309760 creator A5054966819 @default.
• W3213309760 date "2021-11-08" @default.
• W3213309760 modified "2023-09-23" @default.
• W3213309760 title "Near-Optimal Statistical Query Hardness of Learning Halfspaces with Massart Noise." @default.
• W3213309760 cites W1517526469 @default.
• W3213309760 cites W1556014900 @default.
• W3213309760 cites W1601740268 @default.
• W3213309760 cites W1891181203 @default.
• W3213309760 cites W1970543914 @default.
• W3213309760 cites W1971718861 @default.
• W3213309760 cites W1982381767 @default.
• W3213309760 cites W1988790447 @default.
• W3213309760 cites W1990232821 @default.
• W3213309760 cites W1995897489 @default.
• W3213309760 cites W2019363670 @default.
• W3213309760 cites W2026103826 @default.
• W3213309760 cites W2040870580 @default.
• W3213309760 cites W2042038477 @default.
• W3213309760 cites W2064680241 @default.
• W3213309760 cites W2064684330 @default.
• W3213309760 cites W2065151783 @default.
• W3213309760 cites W2065563937 @default.
• W3213309760 cites W2069971337 @default.
• W3213309760 cites W2084544490 @default.
• W3213309760 cites W2086789740 @default.
• W3213309760 cites W2097886040 @default.
• W3213309760 cites W2107189314 @default.
• W3213309760 cites W2109722477 @default.
• W3213309760 cites W2116794057 @default.
• W3213309760 cites W2135933132 @default.
• W3213309760 cites W2140125566 @default.
• W3213309760 cites W2148603752 @default.
• W3213309760 cites W2148858749 @default.
• W3213309760 cites W2155604181 @default.
• W3213309760 cites W2161134697 @default.
• W3213309760 cites W2171574551 @default.
• W3213309760 cites W2256202 @default.
• W3213309760 cites W2475016247 @default.
• W3213309760 cites W2491118 @default.
• W3213309760 cites W2620857485 @default.
• W3213309760 cites W2962703558 @default.
• W3213309760 cites W2962856918 @default.
• W3213309760 cites W2962891378 @default.
• W3213309760 cites W2963351358 @default.
• W3213309760 cites W2963354149 @default.
• W3213309760 cites W2963589833 @default.
• W3213309760 cites W2963595764 @default.
• W3213309760 cites W2963874210 @default.
• W3213309760 cites W2963945243 @default.
• W3213309760 cites W2964008913 @default.
• W3213309760 cites W2964013305 @default.
• W3213309760 cites W2964067168 @default.
• W3213309760 cites W2964135521 @default.
• W3213309760 cites W2964164735 @default.
• W3213309760 cites W2970584880 @default.
• W3213309760 cites W2984272935 @default.
• W3213309760 cites W3006151220 @default.
• W3213309760 cites W3033838839 @default.
• W3213309760 cites W3037262551 @default.
• W3213309760 cites W3046237351 @default.
• W3213309760 cites W3084623378 @default.
• W3213309760 cites W3100344698 @default.
• W3213309760 cites W3112170761 @default.
• W3213309760 cites W3175974551 @default.
• W3213309760 cites W3177410885 @default.
• W3213309760 cites W5594912 @default.
• W3213309760 cites W3184572569 @default.
• W3213309760 hasPublicationYear "2021" @default.
• W3213309760 type Work @default.
• W3213309760 sameAs 3213309760 @default.
• W3213309760 citedByCount "0" @default.
• W3213309760 crossrefType "posted-content" @default.
• W3213309760 hasAuthorship W3213309760A5035962853 @default.
• W3213309760 hasAuthorship W3213309760A5054966819 @default.
• W3213309760 hasConcept C110121322 @default.
• W3213309760 hasConcept C11413529 @default.
• W3213309760 hasConcept C114614502 @default.
• W3213309760 hasConcept C115961682 @default.
• W3213309760 hasConcept C121332964 @default.
• W3213309760 hasConcept C134306372 @default.
• W3213309760 hasConcept C154945302 @default.
• W3213309760 hasConcept C2779557605 @default.
• W3213309760 hasConcept C33923547 @default.
• W3213309760 hasConcept C41008148 @default.
• W3213309760 hasConcept C62520636 @default.
• W3213309760 hasConcept C77553402 @default.
• W3213309760 hasConcept C90119067 @default.
• W3213309760 hasConcept C99498987 @default.
• W3213309760 hasConceptScore W3213309760C110121322 @default.
• W3213309760 hasConceptScore W3213309760C11413529 @default.
• W3213309760 hasConceptScore W3213309760C114614502 @default.
• W3213309760 hasConceptScore W3213309760C115961682 @default.
• W3213309760 hasConceptScore W3213309760C121332964 @default.
• W3213309760 hasConceptScore W3213309760C134306372 @default.
• W3213309760 hasConceptScore W3213309760C154945302 @default.
• W3213309760 hasConceptScore W3213309760C2779557605 @default.
• W3213309760 hasConceptScore W3213309760C33923547 @default.

• next