FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2772097108> ?p ?o ?g. }

• W2772097108 abstract "Author(s): Karianakis, Nikolaos | Advisor(s): Soatto, Stefano | Abstract: Convolutional neural networks (CNNs) have risen to be the de-facto paragon for detecting the presence of objects in a scene, as portrayed by an image. CNNs are described as being approximately to nuisance transformations such as planar translation, both by virtue of their convolutional architecture and by virtue of their approximation properties that, given sufficient parameters and training data, could in principle yield discriminants that are insensitive to nuisance transformations of the data. The fact that contemporary deep convolutional architectures appear very effective in classifying images as containing a given object regardless of its position, scale, and aspect ratio in large-scale benchmarks suggests that the network can effectively manage such nuisance variability. We conduct an empirical study and show that, contrary to popular belief, at the current level of complexity of convolutional architectures and scale of the data sets used to train them, CNNs are not very effective at marginalizing nuisance variability.This discovery leaves researchers the choice of investing more effort in the design of models that are less sensitive to nuisances or designing better region proposal algorithms in an effort to predict where the objects of interest lie and center the model around these regions. In this thesis steps towards both directions are made. First, we introduce DSP-CNN, which deploys domain-size pooling in order to transform the neural networks to be scale invariant in the convolutional operator level. Second, motivated by our empirical analysis, we propose novel sampling and pruning techniques for region proposal schemes that improve the end-to-end performance in large-scale classification, detection and wide-baseline correspondence to state-of-the-art levels. Additionally,since a proposal algorithm involves the design of a classifier, whose results are to be fed to another classifier (a Category CNN), it seems natural to leverage on the latter to design the former. Thus, we introduce a method that leverages on filters learned in the lower layers of CNNs to design a binary boosting classifier for generating class-agnostic proposals. Finally, we extend sampling over time by designing a temporal, hard-attention layer which is trained with reinforcement learning, with application in video sequences for person re-identification." @default.
• W2772097108 created "2017-12-22" @default.
• W2772097108 creator A5076846509 @default.
• W2772097108 date "2017-01-01" @default.
• W2772097108 modified "2023-09-27" @default.
• W2772097108 title "Sampling Algorithms to Handle Nuisances in Large-Scale Recognition" @default.
• W2772097108 hasPublicationYear "2017" @default.
• W2772097108 type Work @default.
• W2772097108 sameAs 2772097108 @default.
• W2772097108 citedByCount "0" @default.
• W2772097108 crossrefType "journal-article" @default.
• W2772097108 hasAuthorship W2772097108A5076846509 @default.
• W2772097108 hasConcept C11413529 @default.
• W2772097108 hasConcept C119857082 @default.
• W2772097108 hasConcept C153180895 @default.
• W2772097108 hasConcept C154945302 @default.
• W2772097108 hasConcept C205649164 @default.
• W2772097108 hasConcept C2778755073 @default.
• W2772097108 hasConcept C41008148 @default.
• W2772097108 hasConcept C58640448 @default.
• W2772097108 hasConcept C70437156 @default.
• W2772097108 hasConcept C81363708 @default.
• W2772097108 hasConceptScore W2772097108C11413529 @default.
• W2772097108 hasConceptScore W2772097108C119857082 @default.
• W2772097108 hasConceptScore W2772097108C153180895 @default.
• W2772097108 hasConceptScore W2772097108C154945302 @default.
• W2772097108 hasConceptScore W2772097108C205649164 @default.
• W2772097108 hasConceptScore W2772097108C2778755073 @default.
• W2772097108 hasConceptScore W2772097108C41008148 @default.
• W2772097108 hasConceptScore W2772097108C58640448 @default.
• W2772097108 hasConceptScore W2772097108C70437156 @default.
• W2772097108 hasConceptScore W2772097108C81363708 @default.
• W2772097108 hasLocation W27720971081 @default.
• W2772097108 hasOpenAccess W2772097108 @default.
• W2772097108 hasPrimaryLocation W27720971081 @default.
• W2772097108 hasRelatedWork W2160684493 @default.
• W2772097108 hasRelatedWork W2234895705 @default.
• W2772097108 hasRelatedWork W2290300449 @default.
• W2772097108 hasRelatedWork W2565756288 @default.
• W2772097108 hasRelatedWork W2592563856 @default.
• W2772097108 hasRelatedWork W2622213512 @default.
• W2772097108 hasRelatedWork W2781459369 @default.
• W2772097108 hasRelatedWork W2791538177 @default.
• W2772097108 hasRelatedWork W2888399003 @default.
• W2772097108 hasRelatedWork W2900768265 @default.
• W2772097108 hasRelatedWork W2952025147 @default.
• W2772097108 hasRelatedWork W2964231884 @default.
• W2772097108 hasRelatedWork W2973039928 @default.
• W2772097108 hasRelatedWork W2981335848 @default.
• W2772097108 hasRelatedWork W3005208096 @default.
• W2772097108 hasRelatedWork W3042888422 @default.
• W2772097108 hasRelatedWork W3105979354 @default.
• W2772097108 hasRelatedWork W3114898850 @default.
• W2772097108 hasRelatedWork W3134609032 @default.
• W2772097108 hasRelatedWork W2147297298 @default.
• W2772097108 isParatext "false" @default.
• W2772097108 isRetracted "false" @default.
• W2772097108 magId "2772097108" @default.
• W2772097108 workType "article" @default.