FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3132155246> ?p ?o ?g. }

• W3132155246 abstract "One of the most fundamental aspects of any machine learning algorithm is the training data used by the algorithm. We introduce the novel concept of ϵ-approximation of datasets, obtaining datasets which are much smaller than or are significant corruptions of the original training data while maintaining similar performance. We introduce a meta-learning algorithm Kernel Inducing Points (KIP) for obtaining such remarkable datasets, drawing inspiration from recent developments in the correspondence between infinitely-wide neural networks and kernel ridge-regression (KRR). For KRR tasks, we demonstrate that KIP can compress datasets by one or two orders of magnitude, significantly improving previous dataset distillation and subset selection methods while obtaining state of the art results for MNIST and CIFAR10 classification. Furthermore, our KIP-learned datasets are transferable to the training of finite-width neural networks even beyond the lazy-training regime. Consequently, we obtain state of the art results for neural network dataset distillation with potential applications to privacy-preservation." @default.
• W3132155246 created "2021-03-01" @default.
• W3132155246 creator A5018500051 @default.
• W3132155246 creator A5064407466 @default.
• W3132155246 creator A5076404171 @default.
• W3132155246 date "2021-05-03" @default.
• W3132155246 modified "2023-09-24" @default.
• W3132155246 title "Dataset Meta-Learning from Kernel-Ridge Regression" @default.
• W3132155246 cites W137285897 @default.
• W3132155246 cites W1605479404 @default.
• W3132155246 cites W2075066642 @default.
• W3132155246 cites W2099768828 @default.
• W3132155246 cites W2112545207 @default.
• W3132155246 cites W2112796928 @default.
• W3132155246 cites W2123872146 @default.
• W3132155246 cites W2135106139 @default.
• W3132155246 cites W2160840682 @default.
• W3132155246 cites W2230030897 @default.
• W3132155246 cites W2401231614 @default.
• W3132155246 cites W2473418344 @default.
• W3132155246 cites W2601450892 @default.
• W3132155246 cites W2809090039 @default.
• W3132155246 cites W2910655610 @default.
• W3132155246 cites W2949746412 @default.
• W3132155246 cites W2962685794 @default.
• W3132155246 cites W2963233958 @default.
• W3132155246 cites W2963323437 @default.
• W3132155246 cites W2963341924 @default.
• W3132155246 cites W2964052793 @default.
• W3132155246 cites W2969656782 @default.
• W3132155246 cites W2971043187 @default.
• W3132155246 cites W2980536810 @default.
• W3132155246 cites W2983647115 @default.
• W3132155246 cites W2994747787 @default.
• W3132155246 cites W2994872659 @default.
• W3132155246 cites W3002073503 @default.
• W3132155246 cites W3032949543 @default.
• W3132155246 cites W3035433747 @default.
• W3132155246 cites W3035559424 @default.
• W3132155246 cites W3037120332 @default.
• W3132155246 cites W3101069636 @default.
• W3132155246 cites W3118146262 @default.
• W3132155246 cites W3118608800 @default.
• W3132155246 cites W3119342105 @default.
• W3132155246 hasPublicationYear "2021" @default.
• W3132155246 type Work @default.
• W3132155246 sameAs 3132155246 @default.
• W3132155246 citedByCount "2" @default.
• W3132155246 countsByYear W31321552462021 @default.
• W3132155246 crossrefType "proceedings-article" @default.
• W3132155246 hasAuthorship W3132155246A5018500051 @default.
• W3132155246 hasAuthorship W3132155246A5064407466 @default.
• W3132155246 hasAuthorship W3132155246A5076404171 @default.
• W3132155246 hasConcept C105795698 @default.
• W3132155246 hasConcept C114614502 @default.
• W3132155246 hasConcept C115903868 @default.
• W3132155246 hasConcept C119857082 @default.
• W3132155246 hasConcept C122280245 @default.
• W3132155246 hasConcept C12267149 @default.
• W3132155246 hasConcept C124101348 @default.
• W3132155246 hasConcept C151730666 @default.
• W3132155246 hasConcept C153180895 @default.
• W3132155246 hasConcept C154945302 @default.
• W3132155246 hasConcept C162324750 @default.
• W3132155246 hasConcept C187736073 @default.
• W3132155246 hasConcept C190502265 @default.
• W3132155246 hasConcept C2780451532 @default.
• W3132155246 hasConcept C2781002164 @default.
• W3132155246 hasConcept C32277403 @default.
• W3132155246 hasConcept C33923547 @default.
• W3132155246 hasConcept C41008148 @default.
• W3132155246 hasConcept C50644808 @default.
• W3132155246 hasConcept C55166926 @default.
• W3132155246 hasConcept C74193536 @default.
• W3132155246 hasConcept C83546350 @default.
• W3132155246 hasConcept C86803240 @default.
• W3132155246 hasConceptScore W3132155246C105795698 @default.
• W3132155246 hasConceptScore W3132155246C114614502 @default.
• W3132155246 hasConceptScore W3132155246C115903868 @default.
• W3132155246 hasConceptScore W3132155246C119857082 @default.
• W3132155246 hasConceptScore W3132155246C122280245 @default.
• W3132155246 hasConceptScore W3132155246C12267149 @default.
• W3132155246 hasConceptScore W3132155246C124101348 @default.
• W3132155246 hasConceptScore W3132155246C151730666 @default.
• W3132155246 hasConceptScore W3132155246C153180895 @default.
• W3132155246 hasConceptScore W3132155246C154945302 @default.
• W3132155246 hasConceptScore W3132155246C162324750 @default.
• W3132155246 hasConceptScore W3132155246C187736073 @default.
• W3132155246 hasConceptScore W3132155246C190502265 @default.
• W3132155246 hasConceptScore W3132155246C2780451532 @default.
• W3132155246 hasConceptScore W3132155246C2781002164 @default.
• W3132155246 hasConceptScore W3132155246C32277403 @default.
• W3132155246 hasConceptScore W3132155246C33923547 @default.
• W3132155246 hasConceptScore W3132155246C41008148 @default.
• W3132155246 hasConceptScore W3132155246C50644808 @default.
• W3132155246 hasConceptScore W3132155246C55166926 @default.
• W3132155246 hasConceptScore W3132155246C74193536 @default.
• W3132155246 hasConceptScore W3132155246C83546350 @default.
• W3132155246 hasConceptScore W3132155246C86803240 @default.
• W3132155246 hasLocation W31321552461 @default.

• next