FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3158067133> ?p ?o ?g. }

• W3158067133 abstract "In this paper, we propose a novel ensembling technique for deep neural networks, which is able to drastically reduce the required memory compared to alternative approaches. In particular, we propose to extract multiple sub-networks from a single, untrained neural network by solving an end-to-end optimization task combining differentiable scaling over the original architecture, with multiple regularization terms favouring the diversity of the ensemble. Since our proposal aims to detect and extract sub-structures, we call it Structured Ensemble. On a large experimental evaluation, we show that our method can achieve higher or comparable accuracy to competing methods while requiring significantly less storage. In addition, we evaluate our ensembles in terms of predictive calibration and uncertainty, showing they compare favourably with the state-of-the-art. Finally, we draw a link with the continual learning literature, and we propose a modification of our framework to handle continuous streams of tasks with a sub-linear memory cost. We compare with a number of alternative strategies to mitigate catastrophic forgetting, highlighting advantages in terms of average accuracy and memory." @default.
• W3158067133 created "2021-05-10" @default.
• W3158067133 creator A5022153057 @default.
• W3158067133 creator A5039283820 @default.
• W3158067133 creator A5050262749 @default.
• W3158067133 date "2021-05-06" @default.
• W3158067133 modified "2023-10-03" @default.
• W3158067133 title "Structured Ensembles: an Approach to Reduce the Memory Footprint of Ensemble Methods" @default.
• W3158067133 cites W1522301498 @default.
• W3158067133 cites W1622484071 @default.
• W3158067133 cites W1682403713 @default.
• W3158067133 cites W1686810756 @default.
• W3158067133 cites W1821462560 @default.
• W3158067133 cites W2060277733 @default.
• W3158067133 cites W2098824882 @default.
• W3158067133 cites W2108598243 @default.
• W3158067133 cites W2126204609 @default.
• W3158067133 cites W2147800946 @default.
• W3158067133 cites W2172734211 @default.
• W3158067133 cites W2194775991 @default.
• W3158067133 cites W2212660284 @default.
• W3158067133 cites W2294370754 @default.
• W3158067133 cites W2515385951 @default.
• W3158067133 cites W2788388592 @default.
• W3158067133 cites W2805003733 @default.
• W3158067133 cites W2898892019 @default.
• W3158067133 cites W2912934387 @default.
• W3158067133 cites W2922466325 @default.
• W3158067133 cites W2939137134 @default.
• W3158067133 cites W2951696358 @default.
• W3158067133 cites W2962724315 @default.
• W3158067133 cites W2963060032 @default.
• W3158067133 cites W2963207607 @default.
• W3158067133 cites W2963238274 @default.
• W3158067133 cites W2963247446 @default.
• W3158067133 cites W2963363373 @default.
• W3158067133 cites W2963384892 @default.
• W3158067133 cites W2963446085 @default.
• W3158067133 cites W2964059111 @default.
• W3158067133 cites W2964189064 @default.
• W3158067133 cites W2964212410 @default.
• W3158067133 cites W2965563166 @default.
• W3158067133 cites W2970859221 @default.
• W3158067133 cites W2992525328 @default.
• W3158067133 cites W2994797252 @default.
• W3158067133 cites W3034234149 @default.
• W3158067133 cites W3035711539 @default.
• W3158067133 cites W3105124822 @default.
• W3158067133 cites W3113303810 @default.
• W3158067133 cites W3118608800 @default.
• W3158067133 cites W3126837587 @default.
• W3158067133 cites W2426267443 @default.
• W3158067133 hasPublicationYear "2021" @default.
• W3158067133 type Work @default.
• W3158067133 sameAs 3158067133 @default.
• W3158067133 citedByCount "0" @default.
• W3158067133 crossrefType "posted-content" @default.
• W3158067133 hasAuthorship W3158067133A5022153057 @default.
• W3158067133 hasAuthorship W3158067133A5039283820 @default.
• W3158067133 hasAuthorship W3158067133A5050262749 @default.
• W3158067133 hasConcept C111919701 @default.
• W3158067133 hasConcept C119857082 @default.
• W3158067133 hasConcept C134306372 @default.
• W3158067133 hasConcept C138885662 @default.
• W3158067133 hasConcept C147764199 @default.
• W3158067133 hasConcept C154945302 @default.
• W3158067133 hasConcept C199360897 @default.
• W3158067133 hasConcept C202615002 @default.
• W3158067133 hasConcept C2776135515 @default.
• W3158067133 hasConcept C33923547 @default.
• W3158067133 hasConcept C41008148 @default.
• W3158067133 hasConcept C41895202 @default.
• W3158067133 hasConcept C45942800 @default.
• W3158067133 hasConcept C50644808 @default.
• W3158067133 hasConcept C7149132 @default.
• W3158067133 hasConcept C74912251 @default.
• W3158067133 hasConceptScore W3158067133C111919701 @default.
• W3158067133 hasConceptScore W3158067133C119857082 @default.
• W3158067133 hasConceptScore W3158067133C134306372 @default.
• W3158067133 hasConceptScore W3158067133C138885662 @default.
• W3158067133 hasConceptScore W3158067133C147764199 @default.
• W3158067133 hasConceptScore W3158067133C154945302 @default.
• W3158067133 hasConceptScore W3158067133C199360897 @default.
• W3158067133 hasConceptScore W3158067133C202615002 @default.
• W3158067133 hasConceptScore W3158067133C2776135515 @default.
• W3158067133 hasConceptScore W3158067133C33923547 @default.
• W3158067133 hasConceptScore W3158067133C41008148 @default.
• W3158067133 hasConceptScore W3158067133C41895202 @default.
• W3158067133 hasConceptScore W3158067133C45942800 @default.
• W3158067133 hasConceptScore W3158067133C50644808 @default.
• W3158067133 hasConceptScore W3158067133C7149132 @default.
• W3158067133 hasConceptScore W3158067133C74912251 @default.
• W3158067133 hasLocation W31580671331 @default.
• W3158067133 hasOpenAccess W3158067133 @default.
• W3158067133 hasPrimaryLocation W31580671331 @default.
• W3158067133 hasRelatedWork W2113145584 @default.
• W3158067133 hasRelatedWork W2182259890 @default.
• W3158067133 hasRelatedWork W2292365397 @default.
• W3158067133 hasRelatedWork W2339855950 @default.
• W3158067133 hasRelatedWork W2415062102 @default.

• next