FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2913880062> ?p ?o ?g. }

• W2913880062 abstract "Two things seem to be indisputable in the contemporary deep learning discourse: 1. The categorical cross-entropy loss after softmax activation is the method of choice for classification. 2. Training a CNN classifier from scratch on small datasets does not work well. In contrast to this, we show that the cosine loss function provides significantly better performance than cross-entropy on datasets with only a handful of samples per class. For example, the accuracy achieved on the CUB-200-2011 dataset without pre-training is by 30% higher than with the cross-entropy loss. Further experiments on other popular datasets confirm our findings. Moreover, we demonstrate that integrating prior knowledge in the form of class hierarchies is straightforward with the cosine loss and improves classification performance further." @default.
• W2913880062 created "2019-02-21" @default.
• W2913880062 creator A5024934744 @default.
• W2913880062 creator A5087237227 @default.
• W2913880062 date "2019-01-25" @default.
• W2913880062 modified "2023-09-27" @default.
• W2913880062 title "Deep Learning on Small Datasets without Pre-Training using Cosine Loss" @default.
• W2913880062 cites W1672197616 @default.
• W2913880062 cites W1797268635 @default.
• W2913880062 cites W1815076433 @default.
• W2913880062 cites W1849277567 @default.
• W2913880062 cites W1954152232 @default.
• W2913880062 cites W2049750909 @default.
• W2913880062 cites W2102605133 @default.
• W2913880062 cites W2103179193 @default.
• W2913880062 cites W2104657103 @default.
• W2913880062 cites W2108598243 @default.
• W2913880062 cites W2117539524 @default.
• W2913880062 cites W2138011018 @default.
• W2913880062 cites W2145607950 @default.
• W2913880062 cites W2152161678 @default.
• W2913880062 cites W2163605009 @default.
• W2913880062 cites W2172140247 @default.
• W2913880062 cites W2183341477 @default.
• W2913880062 cites W2194321275 @default.
• W2913880062 cites W2194775991 @default.
• W2913880062 cites W2250539671 @default.
• W2913880062 cites W2311523351 @default.
• W2913880062 cites W2526145926 @default.
• W2913880062 cites W2533598788 @default.
• W2913880062 cites W2592929672 @default.
• W2913880062 cites W2601450892 @default.
• W2913880062 cites W2625674597 @default.
• W2913880062 cites W2737041163 @default.
• W2913880062 cites W2742093937 @default.
• W2913880062 cites W2747685395 @default.
• W2913880062 cites W2773003563 @default.
• W2913880062 cites W2787247660 @default.
• W2913880062 cites W2798381792 @default.
• W2913880062 cites W2885201931 @default.
• W2913880062 cites W2902332991 @default.
• W2913880062 cites W2905040722 @default.
• W2913880062 cites W2907258563 @default.
• W2913880062 cites W2952305675 @default.
• W2913880062 cites W2962798326 @default.
• W2913880062 cites W2962843773 @default.
• W2913880062 cites W2962898354 @default.
• W2913880062 cites W2963012544 @default.
• W2913880062 cites W2963263347 @default.
• W2913880062 cites W2963341924 @default.
• W2913880062 cites W2963341956 @default.
• W2913880062 cites W2963466847 @default.
• W2913880062 cites W2963709863 @default.
• W2913880062 cites W2964105864 @default.
• W2913880062 cites W3010290582 @default.
• W2913880062 cites W3118608800 @default.
• W2913880062 cites W3137695714 @default.
• W2913880062 cites W2600537992 @default.
• W2913880062 cites W2954574047 @default.
• W2913880062 hasPublicationYear "2019" @default.
• W2913880062 type Work @default.
• W2913880062 sameAs 2913880062 @default.
• W2913880062 citedByCount "4" @default.
• W2913880062 countsByYear W29138800622019 @default.
• W2913880062 countsByYear W29138800622020 @default.
• W2913880062 crossrefType "posted-content" @default.
• W2913880062 hasAuthorship W2913880062A5024934744 @default.
• W2913880062 hasAuthorship W2913880062A5087237227 @default.
• W2913880062 hasConcept C106301342 @default.
• W2913880062 hasConcept C108583219 @default.
• W2913880062 hasConcept C119857082 @default.
• W2913880062 hasConcept C121332964 @default.
• W2913880062 hasConcept C153180895 @default.
• W2913880062 hasConcept C154945302 @default.
• W2913880062 hasConcept C167981619 @default.
• W2913880062 hasConcept C178009071 @default.
• W2913880062 hasConcept C188441871 @default.
• W2913880062 hasConcept C2524010 @default.
• W2913880062 hasConcept C33923547 @default.
• W2913880062 hasConcept C41008148 @default.
• W2913880062 hasConcept C51632099 @default.
• W2913880062 hasConcept C5274069 @default.
• W2913880062 hasConcept C62520636 @default.
• W2913880062 hasConcept C95623464 @default.
• W2913880062 hasConceptScore W2913880062C106301342 @default.
• W2913880062 hasConceptScore W2913880062C108583219 @default.
• W2913880062 hasConceptScore W2913880062C119857082 @default.
• W2913880062 hasConceptScore W2913880062C121332964 @default.
• W2913880062 hasConceptScore W2913880062C153180895 @default.
• W2913880062 hasConceptScore W2913880062C154945302 @default.
• W2913880062 hasConceptScore W2913880062C167981619 @default.
• W2913880062 hasConceptScore W2913880062C178009071 @default.
• W2913880062 hasConceptScore W2913880062C188441871 @default.
• W2913880062 hasConceptScore W2913880062C2524010 @default.
• W2913880062 hasConceptScore W2913880062C33923547 @default.
• W2913880062 hasConceptScore W2913880062C41008148 @default.
• W2913880062 hasConceptScore W2913880062C51632099 @default.
• W2913880062 hasConceptScore W2913880062C5274069 @default.
• W2913880062 hasConceptScore W2913880062C62520636 @default.
• W2913880062 hasConceptScore W2913880062C95623464 @default.

• next