FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W33870844> ?p ?o ?g. }

• W33870844 endingPage "126" @default.
• W33870844 startingPage "118" @default.
• W33870844 abstract "Our objective is to develop formulations and algorithms for efficiently computing the feature selection path - i.e. the variation in classification accuracy as the fraction of selected features is varied from null to unity. Multiple Kernel Learning subject to lp≤1 regularization (lp-MKL) has been demonstrated to be one of the most effective techniques for non-linear feature selection. However, state-of-the-art lp-MKL algorithms are too computationally expensive to be invoked thousands of times to determine the entire path.We propose a novel conjecture which states that, for certain lp-MKL formulations, the number of features selected in the optimal solution monotonically decreases as p is decreased from an initial value to unity. We prove the conjecture, for a generic family of kernel target alignment based formulations, and show that the feature weights themselves decay (grow) monotonically once they are below (above) a certain threshold at optimality. This allows us to develop a path following algorithm that systematically generates optimal feature sets of decreasing size. The proposed algorithm sets certain feature weights directly to zero for potentially large intervals of p thereby reducing optimization costs while simultaneously providing approximation guarantees.We empirically demonstrate that our formulation can lead to classification accuracies which are as much as 10% higher on benchmark data sets not only as compared to other lp-MKL formulations and uniform kernel baselines but also leading feature selection methods. We further demonstrate that our algorithm reduces training time significantly over other path following algorithms and state-of-the-art lp-MKL optimizers such as SMO-MKL. In particular, we generate the entire feature selection path for data sets with a hundred thousand features in approximately half an hour on standard hardware. Entire path generation for such data set is well beyond the scaling capabilities of other methods." @default.
• W33870844 created "2016-06-24" @default.
• W33870844 creator A5051880496 @default.
• W33870844 creator A5067968767 @default.
• W33870844 creator A5086696547 @default.
• W33870844 date "2014-06-21" @default.
• W33870844 modified "2023-10-03" @default.
• W33870844 title "On p-norm Path Following in Multiple Kernel Learning for Non-linear Feature Selection" @default.
• W33870844 cites W1568307856 @default.
• W33870844 cites W1646506067 @default.
• W33870844 cites W1914219707 @default.
• W33870844 cites W1920328734 @default.
• W33870844 cites W1922017469 @default.
• W33870844 cites W1983599491 @default.
• W33870844 cites W1984274769 @default.
• W33870844 cites W2020944503 @default.
• W33870844 cites W2055586095 @default.
• W33870844 cites W2063386797 @default.
• W33870844 cites W2096032893 @default.
• W33870844 cites W2096765209 @default.
• W33870844 cites W2097839764 @default.
• W33870844 cites W2098728436 @default.
• W33870844 cites W2100348027 @default.
• W33870844 cites W2100658528 @default.
• W33870844 cites W2113362355 @default.
• W33870844 cites W2116589987 @default.
• W33870844 cites W2120137102 @default.
• W33870844 cites W2121308062 @default.
• W33870844 cites W2122825543 @default.
• W33870844 cites W2127069950 @default.
• W33870844 cites W2127658298 @default.
• W33870844 cites W2128035438 @default.
• W33870844 cites W2130698119 @default.
• W33870844 cites W2132333011 @default.
• W33870844 cites W2133956794 @default.
• W33870844 cites W2133958955 @default.
• W33870844 cites W2139338362 @default.
• W33870844 cites W2142387771 @default.
• W33870844 cites W2143628714 @default.
• W33870844 cites W2145295623 @default.
• W33870844 cites W2146722014 @default.
• W33870844 cites W2147197303 @default.
• W33870844 cites W2153631847 @default.
• W33870844 cites W2154420049 @default.
• W33870844 cites W2156291289 @default.
• W33870844 cites W2158978974 @default.
• W33870844 cites W2159067232 @default.
• W33870844 cites W2164535072 @default.
• W33870844 cites W2169769141 @default.
• W33870844 cites W2171188027 @default.
• W33870844 cites W2249237221 @default.
• W33870844 cites W2538008885 @default.
• W33870844 cites W2913088541 @default.
• W33870844 cites W2951654630 @default.
• W33870844 cites W53188351 @default.
• W33870844 cites W71452225 @default.
• W33870844 cites W985009302 @default.
• W33870844 cites W2362339555 @default.
• W33870844 hasPublicationYear "2014" @default.
• W33870844 type Work @default.
• W33870844 sameAs 33870844 @default.
• W33870844 citedByCount "9" @default.
• W33870844 countsByYear W338708442014 @default.
• W33870844 countsByYear W338708442015 @default.
• W33870844 countsByYear W338708442016 @default.
• W33870844 countsByYear W338708442017 @default.
• W33870844 countsByYear W338708442020 @default.
• W33870844 crossrefType "proceedings-article" @default.
• W33870844 hasAuthorship W33870844A5051880496 @default.
• W33870844 hasAuthorship W33870844A5067968767 @default.
• W33870844 hasAuthorship W33870844A5086696547 @default.
• W33870844 hasConcept C11413529 @default.
• W33870844 hasConcept C114614502 @default.
• W33870844 hasConcept C122280245 @default.
• W33870844 hasConcept C12267149 @default.
• W33870844 hasConcept C126255220 @default.
• W33870844 hasConcept C13280743 @default.
• W33870844 hasConcept C138885662 @default.
• W33870844 hasConcept C148483581 @default.
• W33870844 hasConcept C153180895 @default.
• W33870844 hasConcept C154945302 @default.
• W33870844 hasConcept C17744445 @default.
• W33870844 hasConcept C185798385 @default.
• W33870844 hasConcept C191795146 @default.
• W33870844 hasConcept C199360897 @default.
• W33870844 hasConcept C199539241 @default.
• W33870844 hasConcept C205649164 @default.
• W33870844 hasConcept C2776135515 @default.
• W33870844 hasConcept C2776401178 @default.
• W33870844 hasConcept C2776879701 @default.
• W33870844 hasConcept C2777735758 @default.
• W33870844 hasConcept C33923547 @default.
• W33870844 hasConcept C41008148 @default.
• W33870844 hasConcept C41895202 @default.
• W33870844 hasConcept C74193536 @default.
• W33870844 hasConceptScore W33870844C11413529 @default.
• W33870844 hasConceptScore W33870844C114614502 @default.
• W33870844 hasConceptScore W33870844C122280245 @default.

• next