Matches in SemOpenAlex for { <https://semopenalex.org/work/W3214397680> ?p ?o ?g. }
- W3214397680 abstract "Given a matrix $D$ describing the pairwise dissimilarities of a data set, a common task is to embed the data points into Euclidean space. The classical multidimensional scaling (cMDS) algorithm is a widespread method to do this. However, theoretical analysis of the robustness of the algorithm and an in-depth analysis of its performance on non-Euclidean metrics is lacking. In this paper, we derive a formula, based on the eigenvalues of a matrix obtained from $D$, for the Frobenius norm of the difference between $D$ and the metric $D_{text{cmds}}$ returned by cMDS. This error analysis leads us to the conclusion that when the derived matrix has a significant number of negative eigenvalues, then $|D-D_{text{cmds}}|_F$, after initially decreasing, will eventually increase as we increase the dimension. Hence, counterintuitively, the quality of the embedding degrades as we increase the dimension. We empirically verify that the Frobenius norm increases as we increase the dimension for a variety of non-Euclidean metrics. We also show on several benchmark datasets that this degradation in the embedding results in the classification accuracy of both simple (e.g., 1-nearest neighbor) and complex (e.g., multi-layer neural nets) classifiers decreasing as we increase the embedding dimension. Finally, our analysis leads us to a new efficiently computable algorithm that returns a matrix $D_l$ that is at least as close to the original distances as $D_t$ (the Euclidean metric closest in $ell_2$ distance). While $D_l$ is not metric, when given as input to cMDS instead of $D$, it empirically results in solutions whose distance to $D$ does not increase when we increase the dimension and the classification accuracy degrades less than the cMDS solution." @default.
- W3214397680 created "2021-11-22" @default.
- W3214397680 creator A5039675729 @default.
- W3214397680 creator A5042502878 @default.
- W3214397680 creator A5054913384 @default.
- W3214397680 creator A5058931457 @default.
- W3214397680 date "2021-12-06" @default.
- W3214397680 modified "2023-10-17" @default.
- W3214397680 title "How can classical multidimensional scaling go wrong" @default.
- W3214397680 hasPublicationYear "2021" @default.
- W3214397680 type Work @default.
- W3214397680 sameAs 3214397680 @default.
- W3214397680 citedByCount "0" @default.
- W3214397680 crossrefType "proceedings-article" @default.
- W3214397680 hasAuthorship W3214397680A5039675729 @default.
- W3214397680 hasAuthorship W3214397680A5042502878 @default.
- W3214397680 hasAuthorship W3214397680A5054913384 @default.
- W3214397680 hasAuthorship W3214397680A5058931457 @default.
- W3214397680 hasConcept C104317684 @default.
- W3214397680 hasConcept C105795698 @default.
- W3214397680 hasConcept C111208986 @default.
- W3214397680 hasConcept C11413529 @default.
- W3214397680 hasConcept C114614502 @default.
- W3214397680 hasConcept C118615104 @default.
- W3214397680 hasConcept C120174047 @default.
- W3214397680 hasConcept C121332964 @default.
- W3214397680 hasConcept C129782007 @default.
- W3214397680 hasConcept C154945302 @default.
- W3214397680 hasConcept C158693339 @default.
- W3214397680 hasConcept C162324750 @default.
- W3214397680 hasConcept C171558263 @default.
- W3214397680 hasConcept C176217482 @default.
- W3214397680 hasConcept C184898388 @default.
- W3214397680 hasConcept C185592680 @default.
- W3214397680 hasConcept C186450821 @default.
- W3214397680 hasConcept C198043062 @default.
- W3214397680 hasConcept C21547014 @default.
- W3214397680 hasConcept C23719512 @default.
- W3214397680 hasConcept C2524010 @default.
- W3214397680 hasConcept C33676613 @default.
- W3214397680 hasConcept C33923547 @default.
- W3214397680 hasConcept C41008148 @default.
- W3214397680 hasConcept C41608201 @default.
- W3214397680 hasConcept C55493867 @default.
- W3214397680 hasConcept C62520636 @default.
- W3214397680 hasConcept C63479239 @default.
- W3214397680 hasConcept C70518039 @default.
- W3214397680 hasConcept C79337645 @default.
- W3214397680 hasConcept C91682802 @default.
- W3214397680 hasConcept C92207270 @default.
- W3214397680 hasConcept C99844830 @default.
- W3214397680 hasConceptScore W3214397680C104317684 @default.
- W3214397680 hasConceptScore W3214397680C105795698 @default.
- W3214397680 hasConceptScore W3214397680C111208986 @default.
- W3214397680 hasConceptScore W3214397680C11413529 @default.
- W3214397680 hasConceptScore W3214397680C114614502 @default.
- W3214397680 hasConceptScore W3214397680C118615104 @default.
- W3214397680 hasConceptScore W3214397680C120174047 @default.
- W3214397680 hasConceptScore W3214397680C121332964 @default.
- W3214397680 hasConceptScore W3214397680C129782007 @default.
- W3214397680 hasConceptScore W3214397680C154945302 @default.
- W3214397680 hasConceptScore W3214397680C158693339 @default.
- W3214397680 hasConceptScore W3214397680C162324750 @default.
- W3214397680 hasConceptScore W3214397680C171558263 @default.
- W3214397680 hasConceptScore W3214397680C176217482 @default.
- W3214397680 hasConceptScore W3214397680C184898388 @default.
- W3214397680 hasConceptScore W3214397680C185592680 @default.
- W3214397680 hasConceptScore W3214397680C186450821 @default.
- W3214397680 hasConceptScore W3214397680C198043062 @default.
- W3214397680 hasConceptScore W3214397680C21547014 @default.
- W3214397680 hasConceptScore W3214397680C23719512 @default.
- W3214397680 hasConceptScore W3214397680C2524010 @default.
- W3214397680 hasConceptScore W3214397680C33676613 @default.
- W3214397680 hasConceptScore W3214397680C33923547 @default.
- W3214397680 hasConceptScore W3214397680C41008148 @default.
- W3214397680 hasConceptScore W3214397680C41608201 @default.
- W3214397680 hasConceptScore W3214397680C55493867 @default.
- W3214397680 hasConceptScore W3214397680C62520636 @default.
- W3214397680 hasConceptScore W3214397680C63479239 @default.
- W3214397680 hasConceptScore W3214397680C70518039 @default.
- W3214397680 hasConceptScore W3214397680C79337645 @default.
- W3214397680 hasConceptScore W3214397680C91682802 @default.
- W3214397680 hasConceptScore W3214397680C92207270 @default.
- W3214397680 hasConceptScore W3214397680C99844830 @default.
- W3214397680 hasLocation W32143976801 @default.
- W3214397680 hasOpenAccess W3214397680 @default.
- W3214397680 hasPrimaryLocation W32143976801 @default.
- W3214397680 hasRelatedWork W2062050527 @default.
- W3214397680 hasRelatedWork W2086041184 @default.
- W3214397680 hasRelatedWork W2107485210 @default.
- W3214397680 hasRelatedWork W2278163498 @default.
- W3214397680 hasRelatedWork W2413390986 @default.
- W3214397680 hasRelatedWork W2532708770 @default.
- W3214397680 hasRelatedWork W2555153366 @default.
- W3214397680 hasRelatedWork W2555479603 @default.
- W3214397680 hasRelatedWork W2785833512 @default.
- W3214397680 hasRelatedWork W2884607871 @default.
- W3214397680 hasRelatedWork W2885643357 @default.
- W3214397680 hasRelatedWork W2899774684 @default.
- W3214397680 hasRelatedWork W2902132144 @default.