FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W102915213> ?p ?o ?g. }

• W102915213 abstract "Principal Components Analysis is a standard tool in data analysis, widely used in data-rich fields such as computer vision, data mining, bioinformatics, and econometrics. For a set of vectors in Rn number k < n, the method returns a subspace of dimension k whose average squared distance to that set is as small as possible. Besides saving computation by reducing the dimension, projecting to this subspace can often reveal structure that was hidden in high dimension. This thesis considers several novel extensions of PCA, which provably reveals hidden structure where standard PCA fails to do so. First, we consider Robust PCA, which prevents a few points, possibly corrupted by an adversary, from having a large effect on the analysis. The key idea is to alternate noise removal with projection to a constant fraction of the dimensions. When applied to noisy mixture models, the algorithm finds a subspace that is close to the pair of means that are furthest apart. By choosing and testing random directions in this subspace, the algorithm finds a partitioning hyperplane that does not cut any component and then recurses on the two resulting halfspaces. This strategy yields a learning algorithm for noisy mixtures of log-concave distributions that is only slightly weaker than the noiseless result (Chap. 5). Second, we consider Isotropic PCA, which can go beyond the first two moments in identifying “interesting” directions in data. The algorithm first makes the distribution isotropic through an affine transformation. Then the algorithm reweights the data and computes the resulting first and second moments. In effect, this simulates a non-isotropic distribution, whose moments are sometimes meaningful. In the case of a mixture of Gaussians under a Gaussian reweighting, either the first moment or the direction of maximum second moment can be used to partition the components of the mixture assuming that the components are sufficiently separated. This strategy leads to the first affine-invariant algorithm that can provably learn mixtures of Gaussians in high dimensions, improving significantly on known results (Chap. 6). Thirdly, we define the “Subgraph Parity Tensor” of order r of a graph and reduce the problem of finding planted cliques in random graphs to the problem of finding the top principal component of this tensor (Chapter 7). This extends work by Frieze and Kannan, which considers only third order tensors. The intuition behind the result is that the entries in the block of the tensor corresponding to the clique will all have the same values, while the values in other blocks will be uncorrelated. This forces the top principal component of the tensor to “point” towards the clique. Using a previously known algorithm, the clique can be recovered." @default.
• W102915213 created "2016-06-24" @default.
• W102915213 creator A5053220327 @default.
• W102915213 creator A5085424042 @default.
• W102915213 date "2009-01-01" @default.
• W102915213 modified "2023-10-18" @default.
• W102915213 title "Extensions of principal components analysis" @default.
• W102915213 cites W1519565561 @default.
• W102915213 cites W1554772789 @default.
• W102915213 cites W1574816920 @default.
• W102915213 cites W1575824241 @default.
• W102915213 cites W1605711022 @default.
• W102915213 cites W1766442844 @default.
• W102915213 cites W1772739125 @default.
• W102915213 cites W1796325169 @default.
• W102915213 cites W18426229 @default.
• W102915213 cites W1956647075 @default.
• W102915213 cites W1966451953 @default.
• W102915213 cites W1967134148 @default.
• W102915213 cites W1967780722 @default.
• W102915213 cites W1968998685 @default.
• W102915213 cites W1969015668 @default.
• W102915213 cites W1976238508 @default.
• W102915213 cites W1985123706 @default.
• W102915213 cites W1985792472 @default.
• W102915213 cites W1995854415 @default.
• W102915213 cites W1998058722 @default.
• W102915213 cites W2010601719 @default.
• W102915213 cites W2014562510 @default.
• W102915213 cites W2049633694 @default.
• W102915213 cites W2062307640 @default.
• W102915213 cites W2063392856 @default.
• W102915213 cites W2072773380 @default.
• W102915213 cites W2078758709 @default.
• W102915213 cites W2087038607 @default.
• W102915213 cites W2087788281 @default.
• W102915213 cites W2088164510 @default.
• W102915213 cites W2102773363 @default.
• W102915213 cites W2120358419 @default.
• W102915213 cites W2122860281 @default.
• W102915213 cites W2127218421 @default.
• W102915213 cites W2135346934 @default.
• W102915213 cites W2146756121 @default.
• W102915213 cites W2166602562 @default.
• W102915213 cites W2294798173 @default.
• W102915213 cites W2401610261 @default.
• W102915213 cites W25941316 @default.
• W102915213 cites W2799061466 @default.
• W102915213 cites W658559791 @default.
• W102915213 cites W206359983 @default.
• W102915213 hasPublicationYear "2009" @default.
• W102915213 type Work @default.
• W102915213 sameAs 102915213 @default.
• W102915213 citedByCount "1" @default.
• W102915213 crossrefType "dissertation" @default.
• W102915213 hasAuthorship W102915213A5053220327 @default.
• W102915213 hasAuthorship W102915213A5085424042 @default.
• W102915213 hasConcept C105795698 @default.
• W102915213 hasConcept C11413529 @default.
• W102915213 hasConcept C114614502 @default.
• W102915213 hasConcept C153180895 @default.
• W102915213 hasConcept C154945302 @default.
• W102915213 hasConcept C202444582 @default.
• W102915213 hasConcept C27438332 @default.
• W102915213 hasConcept C32834561 @default.
• W102915213 hasConcept C33676613 @default.
• W102915213 hasConcept C33923547 @default.
• W102915213 hasConcept C41008148 @default.
• W102915213 hasConcept C45374587 @default.
• W102915213 hasConcept C68693459 @default.
• W102915213 hasConcept C92757383 @default.
• W102915213 hasConceptScore W102915213C105795698 @default.
• W102915213 hasConceptScore W102915213C11413529 @default.
• W102915213 hasConceptScore W102915213C114614502 @default.
• W102915213 hasConceptScore W102915213C153180895 @default.
• W102915213 hasConceptScore W102915213C154945302 @default.
• W102915213 hasConceptScore W102915213C202444582 @default.
• W102915213 hasConceptScore W102915213C27438332 @default.
• W102915213 hasConceptScore W102915213C32834561 @default.
• W102915213 hasConceptScore W102915213C33676613 @default.
• W102915213 hasConceptScore W102915213C33923547 @default.
• W102915213 hasConceptScore W102915213C41008148 @default.
• W102915213 hasConceptScore W102915213C45374587 @default.
• W102915213 hasConceptScore W102915213C68693459 @default.
• W102915213 hasConceptScore W102915213C92757383 @default.
• W102915213 hasLocation W1029152131 @default.
• W102915213 hasOpenAccess W102915213 @default.
• W102915213 hasPrimaryLocation W1029152131 @default.
• W102915213 hasRelatedWork W1581291877 @default.
• W102915213 hasRelatedWork W1902701714 @default.
• W102915213 hasRelatedWork W2010072035 @default.
• W102915213 hasRelatedWork W2047064469 @default.
• W102915213 hasRelatedWork W2057880425 @default.
• W102915213 hasRelatedWork W2251834888 @default.
• W102915213 hasRelatedWork W2291808657 @default.
• W102915213 hasRelatedWork W2788305524 @default.
• W102915213 hasRelatedWork W2901482344 @default.
• W102915213 hasRelatedWork W2902121050 @default.
• W102915213 hasRelatedWork W2950012359 @default.
• W102915213 hasRelatedWork W2950248432 @default.

• next