FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3034635087> ?p ?o ?g. }

• W3034635087 abstract "In over two decades of research, the field of dictionary learning has gathered a large collection of successful applications, and theoretical guarantees for model recovery are known only whenever optimization is carried out in the same model class as that of the underlying dictionary. This work characterizes the surprising phenomenon that dictionary recovery can be facilitated by searching over the space of larger over-realized models. This observation is general and independent of the specific dictionary learning algorithm used. We thoroughly demonstrate this observation in practice and provide an analysis of this phenomenon by tying recovery measures to generalization bounds. In particular, we show that model recovery can be upper-bounded by the empirical risk, a model-dependent quantity and the generalization gap, reflecting our empirical findings. We further show that an efficient and provably correct distillation approach can be employed to recover the correct atoms from the over-realized model. As a result, our meta-algorithm provides dictionary estimates with consistently better recovery of the ground-truth model." @default.
• W3034635087 created "2020-06-19" @default.
• W3034635087 creator A5011989964 @default.
• W3034635087 creator A5064317084 @default.
• W3034635087 creator A5086776097 @default.
• W3034635087 date "2020-06-11" @default.
• W3034635087 modified "2023-09-27" @default.
• W3034635087 title "Recovery and Generalization in Over-Realized Dictionary Learning" @default.
• W3034635087 cites W1591116419 @default.
• W3034635087 cites W1630816465 @default.
• W3034635087 cites W1933990309 @default.
• W3034635087 cites W2024254345 @default.
• W3034635087 cites W2034683677 @default.
• W3034635087 cites W2105459687 @default.
• W3034635087 cites W2105464873 @default.
• W3034635087 cites W2112447569 @default.
• W3034635087 cites W2115429828 @default.
• W3034635087 cites W2116148865 @default.
• W3034635087 cites W2128659236 @default.
• W3034635087 cites W2129131372 @default.
• W3034635087 cites W2134618502 @default.
• W3034635087 cites W2135046866 @default.
• W3034635087 cites W2143238292 @default.
• W3034635087 cites W2146758737 @default.
• W3034635087 cites W2150593711 @default.
• W3034635087 cites W2152046843 @default.
• W3034635087 cites W2154332973 @default.
• W3034635087 cites W2155981690 @default.
• W3034635087 cites W2160547390 @default.
• W3034635087 cites W2163922914 @default.
• W3034635087 cites W2167349346 @default.
• W3034635087 cites W2523452257 @default.
• W3034635087 cites W2557139899 @default.
• W3034635087 cites W2625387573 @default.
• W3034635087 cites W2962959718 @default.
• W3034635087 cites W2963518130 @default.
• W3034635087 cites W2963591837 @default.
• W3034635087 cites W2963641291 @default.
• W3034635087 cites W2963697946 @default.
• W3034635087 cites W2967536008 @default.
• W3034635087 cites W2970540478 @default.
• W3034635087 cites W2975480662 @default.
• W3034635087 cites W2995582708 @default.
• W3034635087 cites W3005710147 @default.
• W3034635087 cites W3008906732 @default.
• W3034635087 cites W3035456141 @default.
• W3034635087 cites W3086393698 @default.
• W3034635087 cites W3099497614 @default.
• W3034635087 cites W3137695714 @default.
• W3034635087 doi "https://doi.org/10.48550/arxiv.2006.06179" @default.
• W3034635087 hasPublicationYear "2020" @default.
• W3034635087 type Work @default.
• W3034635087 sameAs 3034635087 @default.
• W3034635087 citedByCount "1" @default.
• W3034635087 countsByYear W30346350872021 @default.
• W3034635087 crossrefType "posted-content" @default.
• W3034635087 hasAuthorship W3034635087A5011989964 @default.
• W3034635087 hasAuthorship W3034635087A5064317084 @default.
• W3034635087 hasAuthorship W3034635087A5086776097 @default.
• W3034635087 hasBestOaLocation W30346350871 @default.
• W3034635087 hasConcept C105795698 @default.
• W3034635087 hasConcept C111919701 @default.
• W3034635087 hasConcept C11413529 @default.
• W3034635087 hasConcept C119857082 @default.
• W3034635087 hasConcept C120936955 @default.
• W3034635087 hasConcept C121332964 @default.
• W3034635087 hasConcept C127705205 @default.
• W3034635087 hasConcept C134306372 @default.
• W3034635087 hasConcept C154945302 @default.
• W3034635087 hasConcept C177148314 @default.
• W3034635087 hasConcept C2777212361 @default.
• W3034635087 hasConcept C2778572836 @default.
• W3034635087 hasConcept C2780938662 @default.
• W3034635087 hasConcept C33923547 @default.
• W3034635087 hasConcept C34388435 @default.
• W3034635087 hasConcept C41008148 @default.
• W3034635087 hasConcept C50335755 @default.
• W3034635087 hasConcept C62520636 @default.
• W3034635087 hasConcept C80444323 @default.
• W3034635087 hasConceptScore W3034635087C105795698 @default.
• W3034635087 hasConceptScore W3034635087C111919701 @default.
• W3034635087 hasConceptScore W3034635087C11413529 @default.
• W3034635087 hasConceptScore W3034635087C119857082 @default.
• W3034635087 hasConceptScore W3034635087C120936955 @default.
• W3034635087 hasConceptScore W3034635087C121332964 @default.
• W3034635087 hasConceptScore W3034635087C127705205 @default.
• W3034635087 hasConceptScore W3034635087C134306372 @default.
• W3034635087 hasConceptScore W3034635087C154945302 @default.
• W3034635087 hasConceptScore W3034635087C177148314 @default.
• W3034635087 hasConceptScore W3034635087C2777212361 @default.
• W3034635087 hasConceptScore W3034635087C2778572836 @default.
• W3034635087 hasConceptScore W3034635087C2780938662 @default.
• W3034635087 hasConceptScore W3034635087C33923547 @default.
• W3034635087 hasConceptScore W3034635087C34388435 @default.
• W3034635087 hasConceptScore W3034635087C41008148 @default.
• W3034635087 hasConceptScore W3034635087C50335755 @default.
• W3034635087 hasConceptScore W3034635087C62520636 @default.
• W3034635087 hasConceptScore W3034635087C80444323 @default.
• W3034635087 hasLocation W30346350871 @default.
• W3034635087 hasOpenAccess W3034635087 @default.

• next