FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3112091058> ?p ?o ?g. }

• W3112091058 abstract "Classification is one of the major tasks that deep learning is successfully tackling. Categorization is also a fundamental cognitive ability. A well-known perceptual consequence of categorization in humans and other animals, called categorical perception, is characterized by a within-category compression and a between-category separation: two items, close in input space, are perceived closer if they belong to the same category than if they belong to different categories. Elaborating on experimental and theoretical results in cognitive science, here we study categorical effects in artificial neural networks. Our formal and numerical analysis provides insights into the geometry of the neural representation in deep layers, with expansion of space near category boundaries and contraction far from category boundaries. We investigate categorical representation by using two complementary approaches: one mimics experiments in psychophysics and cognitive neuroscience by means of morphed continua between stimuli of different categories, while the other introduces a categoricality index that quantifies the separability of the classes at the population level (a given layer in the neural network). We show on both shallow and deep neural networks that category learning automatically induces categorical perception. We further show that the deeper a layer, the stronger the categorical effects. An important outcome of our analysis is to provide a coherent and unifying view of the efficacy of different heuristic practices of the dropout regularization technique. Our views, which find echoes in the neuroscience literature, insist on the differential role of noise as a function of the level of representation and in the course of learning: noise injected in the hidden layers gets structured according to the organization of the categories, more variability being allowed within a category than across classes." @default.
• W3112091058 created "2020-12-21" @default.
• W3112091058 creator A5044018048 @default.
• W3112091058 creator A5067175746 @default.
• W3112091058 date "2020-12-10" @default.
• W3112091058 modified "2023-09-23" @default.
• W3112091058 title "Categorical Perception: A Groundwork for Deep Learning." @default.
• W3112091058 cites W1496559305 @default.
• W3112091058 cites W1515692101 @default.
• W3112091058 cites W1521968289 @default.
• W3112091058 cites W1560089794 @default.
• W3112091058 cites W1686810756 @default.
• W3112091058 cites W1799366690 @default.
• W3112091058 cites W1974933912 @default.
• W3112091058 cites W1975849605 @default.
• W3112091058 cites W1979482308 @default.
• W3112091058 cites W1994492508 @default.
• W3112091058 cites W2003766373 @default.
• W3112091058 cites W2011301426 @default.
• W3112091058 cites W2015099338 @default.
• W3112091058 cites W2029310539 @default.
• W3112091058 cites W2039396020 @default.
• W3112091058 cites W2042295619 @default.
• W3112091058 cites W2046634051 @default.
• W3112091058 cites W2051275520 @default.
• W3112091058 cites W2051856012 @default.
• W3112091058 cites W2056001434 @default.
• W3112091058 cites W2057853719 @default.
• W3112091058 cites W2058960948 @default.
• W3112091058 cites W2060971741 @default.
• W3112091058 cites W2073526836 @default.
• W3112091058 cites W2076063813 @default.
• W3112091058 cites W2082195896 @default.
• W3112091058 cites W2086396589 @default.
• W3112091058 cites W2090614046 @default.
• W3112091058 cites W2092189137 @default.
• W3112091058 cites W2095705004 @default.
• W3112091058 cites W2098874108 @default.
• W3112091058 cites W2099111195 @default.
• W3112091058 cites W2099471712 @default.
• W3112091058 cites W2101234009 @default.
• W3112091058 cites W2105693527 @default.
• W3112091058 cites W2108598243 @default.
• W3112091058 cites W2111376597 @default.
• W3112091058 cites W2111406701 @default.
• W3112091058 cites W2112796928 @default.
• W3112091058 cites W2117539524 @default.
• W3112091058 cites W2123544472 @default.
• W3112091058 cites W2124136621 @default.
• W3112091058 cites W2124544892 @default.
• W3112091058 cites W2149467628 @default.
• W3112091058 cites W2153094827 @default.
• W3112091058 cites W2154649497 @default.
• W3112091058 cites W2157695255 @default.
• W3112091058 cites W2163605009 @default.
• W3112091058 cites W2164949385 @default.
• W3112091058 cites W2166405697 @default.
• W3112091058 cites W2213467510 @default.
• W3112091058 cites W2296073425 @default.
• W3112091058 cites W2339016247 @default.
• W3112091058 cites W2342249984 @default.
• W3112091058 cites W2557283755 @default.
• W3112091058 cites W2593484309 @default.
• W3112091058 cites W2593634001 @default.
• W3112091058 cites W2781596748 @default.
• W3112091058 cites W2887411036 @default.
• W3112091058 cites W2919115771 @default.
• W3112091058 cites W2923274354 @default.
• W3112091058 cites W2952502547 @default.
• W3112091058 cites W2961637526 @default.
• W3112091058 cites W2963836885 @default.
• W3112091058 cites W2964121744 @default.
• W3112091058 cites W2969418762 @default.
• W3112091058 cites W3099878876 @default.
• W3112091058 cites W3100481436 @default.
• W3112091058 cites W3118438167 @default.
• W3112091058 cites W3118608800 @default.
• W3112091058 cites W3213887908 @default.
• W3112091058 hasPublicationYear "2020" @default.
• W3112091058 type Work @default.
• W3112091058 sameAs 3112091058 @default.
• W3112091058 citedByCount "0" @default.
• W3112091058 crossrefType "posted-content" @default.
• W3112091058 hasAuthorship W3112091058A5044018048 @default.
• W3112091058 hasAuthorship W3112091058A5067175746 @default.
• W3112091058 hasConcept C119857082 @default.
• W3112091058 hasConcept C144024400 @default.
• W3112091058 hasConcept C149923435 @default.
• W3112091058 hasConcept C154945302 @default.
• W3112091058 hasConcept C15744967 @default.
• W3112091058 hasConcept C169760540 @default.
• W3112091058 hasConcept C169900460 @default.
• W3112091058 hasConcept C17289045 @default.
• W3112091058 hasConcept C180747234 @default.
• W3112091058 hasConcept C26760741 @default.
• W3112091058 hasConcept C2908647359 @default.
• W3112091058 hasConcept C33640556 @default.
• W3112091058 hasConcept C41008148 @default.
• W3112091058 hasConcept C48164120 @default.
• W3112091058 hasConcept C50644808 @default.

• next