Matches in SemOpenAlex for { <https://semopenalex.org/work/W3100823189> ?p ?o ?g. }
- W3100823189 endingPage "1620" @default.
- W3100823189 startingPage "1597" @default.
- W3100823189 abstract "We present a class of algorithms capable of directly training deep neural networks with respect to large families of task-specific performance measures such as the F-measure and the Kullback-Leibler divergence that are structured and non-decomposable. This presents a departure from standard deep learning techniques that typically use squared or cross-entropy loss functions (that are decomposable) to train neural networks. We demonstrate that directly training with task-specific loss functions yields much faster and more stable convergence across problems and datasets. Our proposed algorithms and implementations have several novel features including (i) convergence to first order stationary points despite optimizing complex objective functions; (ii) use of fewer training samples to achieve a desired level of convergence, (iii) a substantial reduction in training time, and (iv) a seamless integration of our implementation into existing symbolic gradient frameworks. We implement our techniques on a variety of deep architectures including multi-layer perceptrons and recurrent neural networks and show that on a variety of benchmark and real data sets, our algorithms outperform traditional approaches to training deep networks, as well as some recent approaches to task-specific training of neural networks." @default.
- W3100823189 created "2020-11-23" @default.
- W3100823189 creator A5037947876 @default.
- W3100823189 creator A5050494173 @default.
- W3100823189 creator A5063975186 @default.
- W3100823189 creator A5081725635 @default.
- W3100823189 creator A5084362470 @default.
- W3100823189 date "2018-07-02" @default.
- W3100823189 modified "2023-10-18" @default.
- W3100823189 title "Optimizing non-decomposable measures with deep networks" @default.
- W3100823189 cites W1773633997 @default.
- W3100823189 cites W1919365417 @default.
- W3100823189 cites W2028034626 @default.
- W3100823189 cites W2031248101 @default.
- W3100823189 cites W2067977748 @default.
- W3100823189 cites W2133138357 @default.
- W3100823189 cites W2152175008 @default.
- W3100823189 cites W2186678232 @default.
- W3100823189 cites W2467434233 @default.
- W3100823189 cites W2783218948 @default.
- W3100823189 cites W3099955943 @default.
- W3100823189 cites W4213009331 @default.
- W3100823189 doi "https://doi.org/10.1007/s10994-018-5736-y" @default.
- W3100823189 hasPublicationYear "2018" @default.
- W3100823189 type Work @default.
- W3100823189 sameAs 3100823189 @default.
- W3100823189 citedByCount "26" @default.
- W3100823189 countsByYear W31008231892018 @default.
- W3100823189 countsByYear W31008231892019 @default.
- W3100823189 countsByYear W31008231892020 @default.
- W3100823189 countsByYear W31008231892021 @default.
- W3100823189 countsByYear W31008231892022 @default.
- W3100823189 countsByYear W31008231892023 @default.
- W3100823189 crossrefType "journal-article" @default.
- W3100823189 hasAuthorship W3100823189A5037947876 @default.
- W3100823189 hasAuthorship W3100823189A5050494173 @default.
- W3100823189 hasAuthorship W3100823189A5063975186 @default.
- W3100823189 hasAuthorship W3100823189A5081725635 @default.
- W3100823189 hasAuthorship W3100823189A5084362470 @default.
- W3100823189 hasBestOaLocation W31008231891 @default.
- W3100823189 hasConcept C106301342 @default.
- W3100823189 hasConcept C108583219 @default.
- W3100823189 hasConcept C119857082 @default.
- W3100823189 hasConcept C121332964 @default.
- W3100823189 hasConcept C13280743 @default.
- W3100823189 hasConcept C136197465 @default.
- W3100823189 hasConcept C138885662 @default.
- W3100823189 hasConcept C154945302 @default.
- W3100823189 hasConcept C162324750 @default.
- W3100823189 hasConcept C185798385 @default.
- W3100823189 hasConcept C187736073 @default.
- W3100823189 hasConcept C205649164 @default.
- W3100823189 hasConcept C207390915 @default.
- W3100823189 hasConcept C2777303404 @default.
- W3100823189 hasConcept C2780451532 @default.
- W3100823189 hasConcept C2984842247 @default.
- W3100823189 hasConcept C41008148 @default.
- W3100823189 hasConcept C41895202 @default.
- W3100823189 hasConcept C50522688 @default.
- W3100823189 hasConcept C50644808 @default.
- W3100823189 hasConcept C60908668 @default.
- W3100823189 hasConcept C62520636 @default.
- W3100823189 hasConceptScore W3100823189C106301342 @default.
- W3100823189 hasConceptScore W3100823189C108583219 @default.
- W3100823189 hasConceptScore W3100823189C119857082 @default.
- W3100823189 hasConceptScore W3100823189C121332964 @default.
- W3100823189 hasConceptScore W3100823189C13280743 @default.
- W3100823189 hasConceptScore W3100823189C136197465 @default.
- W3100823189 hasConceptScore W3100823189C138885662 @default.
- W3100823189 hasConceptScore W3100823189C154945302 @default.
- W3100823189 hasConceptScore W3100823189C162324750 @default.
- W3100823189 hasConceptScore W3100823189C185798385 @default.
- W3100823189 hasConceptScore W3100823189C187736073 @default.
- W3100823189 hasConceptScore W3100823189C205649164 @default.
- W3100823189 hasConceptScore W3100823189C207390915 @default.
- W3100823189 hasConceptScore W3100823189C2777303404 @default.
- W3100823189 hasConceptScore W3100823189C2780451532 @default.
- W3100823189 hasConceptScore W3100823189C2984842247 @default.
- W3100823189 hasConceptScore W3100823189C41008148 @default.
- W3100823189 hasConceptScore W3100823189C41895202 @default.
- W3100823189 hasConceptScore W3100823189C50522688 @default.
- W3100823189 hasConceptScore W3100823189C50644808 @default.
- W3100823189 hasConceptScore W3100823189C60908668 @default.
- W3100823189 hasConceptScore W3100823189C62520636 @default.
- W3100823189 hasIssue "8-10" @default.
- W3100823189 hasLocation W31008231891 @default.
- W3100823189 hasLocation W31008231892 @default.
- W3100823189 hasOpenAccess W3100823189 @default.
- W3100823189 hasPrimaryLocation W31008231891 @default.
- W3100823189 hasRelatedWork W1501213224 @default.
- W3100823189 hasRelatedWork W2950066684 @default.
- W3100823189 hasRelatedWork W2991591812 @default.
- W3100823189 hasRelatedWork W3082895349 @default.
- W3100823189 hasRelatedWork W4231994957 @default.
- W3100823189 hasRelatedWork W4288853838 @default.
- W3100823189 hasRelatedWork W4298388782 @default.
- W3100823189 hasRelatedWork W4312831135 @default.
- W3100823189 hasRelatedWork W4322750901 @default.