FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2984884190> ?p ?o ?g. }

• W2984884190 abstract "Deep image embedding aims at learning a convolutional neural network (CNN) based mapping function that maps an image to a feature vector. The embedding quality is usually evaluated by the performance in image search tasks. Since very few users bother to open the second page search results, top-k precision mostly dominates the user experience and thus is one of the crucial evaluation metrics for the embedding quality. Despite being extensively studied, existing algorithms are usually based on heuristic observation without theoretical guarantee. Consequently, gradient descent direction on the training loss is mostly inconsistent with the direction of optimizing the concerned evaluation metric. This inconsistency certainly misleads the training direction and degrades the performance. In contrast to existing works, in this paper, we propose a novel deep image embedding algorithm with end-to-end optimization to top-k precision, the evaluation metric that is closely related to user experience. Specially, our loss function is constructed with wisely selected ``misplaced images along the top k nearest neighbor decision boundary, so that the gradient descent update directly promotes the concerned metric, top-k precision. Further more, our theoretical analysis on the upper bounding and consistency properties of the proposed loss supports that minimizing our proposed loss is equivalent to maximizing top-k precision. Experiments show that our proposed algorithm outperforms all compared state-of-the-art deep image embedding algorithms on three benchmark datasets." @default.
• W2984884190 created "2019-11-22" @default.
• W2984884190 creator A5012371005 @default.
• W2984884190 creator A5017597537 @default.
• W2984884190 creator A5024879728 @default.
• W2984884190 creator A5034651074 @default.
• W2984884190 creator A5060160101 @default.
• W2984884190 date "2019-10-01" @default.
• W2984884190 modified "2023-10-02" @default.
• W2984884190 title "Sampling Wisely: Deep Image Embedding by Top-K Precision Optimization" @default.
• W2984884190 cites W2097117768 @default.
• W2984884190 cites W2102765684 @default.
• W2984884190 cites W2123229215 @default.
• W2984884190 cites W2138011018 @default.
• W2984884190 cites W2300352888 @default.
• W2984884190 cites W2531295232 @default.
• W2984884190 cites W2605102252 @default.
• W2984884190 cites W2613997951 @default.
• W2984884190 cites W2741943936 @default.
• W2984884190 cites W2765467597 @default.
• W2984884190 cites W2779066530 @default.
• W2984884190 cites W2789546350 @default.
• W2984884190 cites W2885563408 @default.
• W2984884190 cites W2895347732 @default.
• W2984884190 cites W2921310091 @default.
• W2984884190 cites W2953271441 @default.
• W2984884190 cites W2962898354 @default.
• W2984884190 cites W2963026686 @default.
• W2984884190 cites W2963350250 @default.
• W2984884190 cites W2963446712 @default.
• W2984884190 cites W2963988212 @default.
• W2984884190 cites W2964015640 @default.
• W2984884190 cites W2964136035 @default.
• W2984884190 cites W3099206234 @default.
• W2984884190 doi "https://doi.org/10.1109/iccv.2019.00805" @default.
• W2984884190 hasPublicationYear "2019" @default.
• W2984884190 type Work @default.
• W2984884190 sameAs 2984884190 @default.
• W2984884190 citedByCount "22" @default.
• W2984884190 countsByYear W29848841902019 @default.
• W2984884190 countsByYear W29848841902020 @default.
• W2984884190 countsByYear W29848841902021 @default.
• W2984884190 countsByYear W29848841902022 @default.
• W2984884190 countsByYear W29848841902023 @default.
• W2984884190 crossrefType "proceedings-article" @default.
• W2984884190 hasAuthorship W2984884190A5012371005 @default.
• W2984884190 hasAuthorship W2984884190A5017597537 @default.
• W2984884190 hasAuthorship W2984884190A5024879728 @default.
• W2984884190 hasAuthorship W2984884190A5034651074 @default.
• W2984884190 hasAuthorship W2984884190A5060160101 @default.
• W2984884190 hasConcept C108583219 @default.
• W2984884190 hasConcept C11413529 @default.
• W2984884190 hasConcept C115961682 @default.
• W2984884190 hasConcept C13280743 @default.
• W2984884190 hasConcept C138885662 @default.
• W2984884190 hasConcept C147037132 @default.
• W2984884190 hasConcept C153258448 @default.
• W2984884190 hasConcept C154945302 @default.
• W2984884190 hasConcept C162324750 @default.
• W2984884190 hasConcept C173801870 @default.
• W2984884190 hasConcept C176217482 @default.
• W2984884190 hasConcept C185798385 @default.
• W2984884190 hasConcept C205649164 @default.
• W2984884190 hasConcept C21547014 @default.
• W2984884190 hasConcept C2776401178 @default.
• W2984884190 hasConcept C41008148 @default.
• W2984884190 hasConcept C41608201 @default.
• W2984884190 hasConcept C41895202 @default.
• W2984884190 hasConcept C50644808 @default.
• W2984884190 hasConcept C55020928 @default.
• W2984884190 hasConcept C63584917 @default.
• W2984884190 hasConcept C81363708 @default.
• W2984884190 hasConceptScore W2984884190C108583219 @default.
• W2984884190 hasConceptScore W2984884190C11413529 @default.
• W2984884190 hasConceptScore W2984884190C115961682 @default.
• W2984884190 hasConceptScore W2984884190C13280743 @default.
• W2984884190 hasConceptScore W2984884190C138885662 @default.
• W2984884190 hasConceptScore W2984884190C147037132 @default.
• W2984884190 hasConceptScore W2984884190C153258448 @default.
• W2984884190 hasConceptScore W2984884190C154945302 @default.
• W2984884190 hasConceptScore W2984884190C162324750 @default.
• W2984884190 hasConceptScore W2984884190C173801870 @default.
• W2984884190 hasConceptScore W2984884190C176217482 @default.
• W2984884190 hasConceptScore W2984884190C185798385 @default.
• W2984884190 hasConceptScore W2984884190C205649164 @default.
• W2984884190 hasConceptScore W2984884190C21547014 @default.
• W2984884190 hasConceptScore W2984884190C2776401178 @default.
• W2984884190 hasConceptScore W2984884190C41008148 @default.
• W2984884190 hasConceptScore W2984884190C41608201 @default.
• W2984884190 hasConceptScore W2984884190C41895202 @default.
• W2984884190 hasConceptScore W2984884190C50644808 @default.
• W2984884190 hasConceptScore W2984884190C55020928 @default.
• W2984884190 hasConceptScore W2984884190C63584917 @default.
• W2984884190 hasConceptScore W2984884190C81363708 @default.
• W2984884190 hasLocation W29848841901 @default.
• W2984884190 hasOpenAccess W2984884190 @default.
• W2984884190 hasPrimaryLocation W29848841901 @default.
• W2984884190 hasRelatedWork W2736037340 @default.
• W2984884190 hasRelatedWork W2756241593 @default.
• W2984884190 hasRelatedWork W2801801420 @default.

• next