FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2912083667> ?p ?o ?g. }

• W2912083667 abstract "We propose a novel image sampling method for differentiable image transformation in deep neural networks. The sampling schemes currently used in deep learning, such as Spatial Transformer Networks, rely on bilinear interpolation, which performs poorly under severe scale changes, and more importantly, results in poor gradient propagation. This is due to their strict reliance on direct neighbors. Instead, we propose to generate random auxiliary samples in the vicinity of each pixel in the sampled image, and create a linear approximation with their intensity values. We then use this approximation as a differentiable formula for the transformed image. We demonstrate that our approach produces more representative gradients with a wider basin of convergence for image alignment, which leads to considerable performance improvements when training networks for classification tasks. This is not only true under large downsampling, but also when there are no scale changes. We compare our approach with multi-scale sampling and show that we outperform it. We then demonstrate that our improvements to the sampler are compatible with other tangential improvements to Spatial Transformer Networks and that it further improves their performance." @default.
• W2912083667 created "2019-02-21" @default.
• W2912083667 creator A5002942261 @default.
• W2912083667 creator A5037094498 @default.
• W2912083667 creator A5049610895 @default.
• W2912083667 creator A5049893251 @default.
• W2912083667 creator A5050009113 @default.
• W2912083667 date "2019-01-21" @default.
• W2912083667 modified "2023-10-16" @default.
• W2912083667 title "Linearized Multi-Sampling for Differentiable Image Transformation" @default.
• W2912083667 cites W1522301498 @default.
• W2912083667 cites W1901129140 @default.
• W2912083667 cites W1903029394 @default.
• W2912083667 cites W2035379092 @default.
• W2912083667 cites W2097073572 @default.
• W2912083667 cites W2097915756 @default.
• W2912083667 cites W2112796928 @default.
• W2912083667 cites W2117876524 @default.
• W2912083667 cites W2118877769 @default.
• W2912083667 cites W2125188192 @default.
• W2912083667 cites W2144348684 @default.
• W2912083667 cites W2155394491 @default.
• W2912083667 cites W2263386426 @default.
• W2912083667 cites W2320444803 @default.
• W2912083667 cites W2340296683 @default.
• W2912083667 cites W2412782625 @default.
• W2912083667 cites W2423984454 @default.
• W2912083667 cites W2520707372 @default.
• W2912083667 cites W2560609797 @default.
• W2912083667 cites W2562066862 @default.
• W2912083667 cites W2601564443 @default.
• W2912083667 cites W2737590896 @default.
• W2912083667 cites W2777357193 @default.
• W2912083667 cites W2781155895 @default.
• W2912083667 cites W2786143498 @default.
• W2912083667 cites W2798450692 @default.
• W2912083667 cites W2884354140 @default.
• W2912083667 cites W2899281149 @default.
• W2912083667 cites W2899771611 @default.
• W2912083667 cites W2963202462 @default.
• W2912083667 cites W2963537932 @default.
• W2912083667 cites W2963564809 @default.
• W2912083667 cites W2963635628 @default.
• W2912083667 cites W2963758027 @default.
• W2912083667 cites W2963857746 @default.
• W2912083667 cites W2981799316 @default.
• W2912083667 cites W603908379 @default.
• W2912083667 doi "https://doi.org/10.48550/arxiv.1901.07124" @default.
• W2912083667 hasPublicationYear "2019" @default.
• W2912083667 type Work @default.
• W2912083667 sameAs 2912083667 @default.
• W2912083667 citedByCount "0" @default.
• W2912083667 crossrefType "posted-content" @default.
• W2912083667 hasAuthorship W2912083667A5002942261 @default.
• W2912083667 hasAuthorship W2912083667A5037094498 @default.
• W2912083667 hasAuthorship W2912083667A5049610895 @default.
• W2912083667 hasAuthorship W2912083667A5049893251 @default.
• W2912083667 hasAuthorship W2912083667A5050009113 @default.
• W2912083667 hasBestOaLocation W29120836671 @default.
• W2912083667 hasConcept C104317684 @default.
• W2912083667 hasConcept C106131492 @default.
• W2912083667 hasConcept C110384440 @default.
• W2912083667 hasConcept C11413529 @default.
• W2912083667 hasConcept C115961682 @default.
• W2912083667 hasConcept C134306372 @default.
• W2912083667 hasConcept C137800194 @default.
• W2912083667 hasConcept C140779682 @default.
• W2912083667 hasConcept C154945302 @default.
• W2912083667 hasConcept C160633673 @default.
• W2912083667 hasConcept C185592680 @default.
• W2912083667 hasConcept C202615002 @default.
• W2912083667 hasConcept C203616005 @default.
• W2912083667 hasConcept C204241405 @default.
• W2912083667 hasConcept C205203396 @default.
• W2912083667 hasConcept C27405340 @default.
• W2912083667 hasConcept C28826006 @default.
• W2912083667 hasConcept C2984842247 @default.
• W2912083667 hasConcept C31972630 @default.
• W2912083667 hasConcept C33923547 @default.
• W2912083667 hasConcept C41008148 @default.
• W2912083667 hasConcept C50644808 @default.
• W2912083667 hasConcept C55493867 @default.
• W2912083667 hasConcept C9417928 @default.
• W2912083667 hasConceptScore W2912083667C104317684 @default.
• W2912083667 hasConceptScore W2912083667C106131492 @default.
• W2912083667 hasConceptScore W2912083667C110384440 @default.
• W2912083667 hasConceptScore W2912083667C11413529 @default.
• W2912083667 hasConceptScore W2912083667C115961682 @default.
• W2912083667 hasConceptScore W2912083667C134306372 @default.
• W2912083667 hasConceptScore W2912083667C137800194 @default.
• W2912083667 hasConceptScore W2912083667C140779682 @default.
• W2912083667 hasConceptScore W2912083667C154945302 @default.
• W2912083667 hasConceptScore W2912083667C160633673 @default.
• W2912083667 hasConceptScore W2912083667C185592680 @default.
• W2912083667 hasConceptScore W2912083667C202615002 @default.
• W2912083667 hasConceptScore W2912083667C203616005 @default.
• W2912083667 hasConceptScore W2912083667C204241405 @default.
• W2912083667 hasConceptScore W2912083667C205203396 @default.
• W2912083667 hasConceptScore W2912083667C27405340 @default.
• W2912083667 hasConceptScore W2912083667C28826006 @default.

• next