Matches in SemOpenAlex for { <https://semopenalex.org/work/W3100477570> ?p ?o ?g. }
- W3100477570 abstract "Feature extraction with convolutional neural networks (CNNs) is a popular method to represent images for machine learning tasks. These representations seek to capture global image content, and ideally should be independent of geometric transformations. We focus on measuring and visualizing the shift invariance of extracted features from popular off-the-shelf CNN models. We present the results of three experiments comparing representations of millions of images with exhaustively shifted objects, examining both local invariance (within a few pixels) and global invariance (across the image frame). We conclude that features extracted from popular networks are not globally invariant, and that biases and artifacts exist within this variance. Additionally, we determine that anti-aliased models significantly improve local invariance but do not impact global invariance. Finally, we provide a code repository for experiment reproduction, as well as a website to interact with our results at this https URL." @default.
- W3100477570 created "2020-11-23" @default.
- W3100477570 creator A5048522863 @default.
- W3100477570 creator A5056288677 @default.
- W3100477570 creator A5064131510 @default.
- W3100477570 date "2020-11-09" @default.
- W3100477570 modified "2023-09-27" @default.
- W3100477570 title "What Does CNN Shift Invariance Look Like? A Visualization Study" @default.
- W3100477570 cites W1484816920 @default.
- W3100477570 cites W1524680991 @default.
- W3100477570 cites W1593727536 @default.
- W3100477570 cites W1861492603 @default.
- W3100477570 cites W2117539524 @default.
- W3100477570 cites W2134670479 @default.
- W3100477570 cites W2163605009 @default.
- W3100477570 cites W2194775991 @default.
- W3100477570 cites W2204975001 @default.
- W3100477570 cites W2773726006 @default.
- W3100477570 cites W2775273147 @default.
- W3100477570 cites W2913314773 @default.
- W3100477570 cites W2954996726 @default.
- W3100477570 cites W2963150697 @default.
- W3100477570 cites W2963163009 @default.
- W3100477570 cites W2963174698 @default.
- W3100477570 cites W2963376432 @default.
- W3100477570 cites W2963684088 @default.
- W3100477570 cites W2967821872 @default.
- W3100477570 cites W2970971581 @default.
- W3100477570 cites W2996904338 @default.
- W3100477570 cites W3035106315 @default.
- W3100477570 cites W603908379 @default.
- W3100477570 hasPublicationYear "2020" @default.
- W3100477570 type Work @default.
- W3100477570 sameAs 3100477570 @default.
- W3100477570 citedByCount "0" @default.
- W3100477570 crossrefType "posted-content" @default.
- W3100477570 hasAuthorship W3100477570A5048522863 @default.
- W3100477570 hasAuthorship W3100477570A5056288677 @default.
- W3100477570 hasAuthorship W3100477570A5064131510 @default.
- W3100477570 hasConcept C108583219 @default.
- W3100477570 hasConcept C115961682 @default.
- W3100477570 hasConcept C121955636 @default.
- W3100477570 hasConcept C126042441 @default.
- W3100477570 hasConcept C144133560 @default.
- W3100477570 hasConcept C153180895 @default.
- W3100477570 hasConcept C154945302 @default.
- W3100477570 hasConcept C160633673 @default.
- W3100477570 hasConcept C177264268 @default.
- W3100477570 hasConcept C190470478 @default.
- W3100477570 hasConcept C196083921 @default.
- W3100477570 hasConcept C199360897 @default.
- W3100477570 hasConcept C2776760102 @default.
- W3100477570 hasConcept C31972630 @default.
- W3100477570 hasConcept C33923547 @default.
- W3100477570 hasConcept C36464697 @default.
- W3100477570 hasConcept C37914503 @default.
- W3100477570 hasConcept C41008148 @default.
- W3100477570 hasConcept C76155785 @default.
- W3100477570 hasConcept C81363708 @default.
- W3100477570 hasConceptScore W3100477570C108583219 @default.
- W3100477570 hasConceptScore W3100477570C115961682 @default.
- W3100477570 hasConceptScore W3100477570C121955636 @default.
- W3100477570 hasConceptScore W3100477570C126042441 @default.
- W3100477570 hasConceptScore W3100477570C144133560 @default.
- W3100477570 hasConceptScore W3100477570C153180895 @default.
- W3100477570 hasConceptScore W3100477570C154945302 @default.
- W3100477570 hasConceptScore W3100477570C160633673 @default.
- W3100477570 hasConceptScore W3100477570C177264268 @default.
- W3100477570 hasConceptScore W3100477570C190470478 @default.
- W3100477570 hasConceptScore W3100477570C196083921 @default.
- W3100477570 hasConceptScore W3100477570C199360897 @default.
- W3100477570 hasConceptScore W3100477570C2776760102 @default.
- W3100477570 hasConceptScore W3100477570C31972630 @default.
- W3100477570 hasConceptScore W3100477570C33923547 @default.
- W3100477570 hasConceptScore W3100477570C36464697 @default.
- W3100477570 hasConceptScore W3100477570C37914503 @default.
- W3100477570 hasConceptScore W3100477570C41008148 @default.
- W3100477570 hasConceptScore W3100477570C76155785 @default.
- W3100477570 hasConceptScore W3100477570C81363708 @default.
- W3100477570 hasLocation W31004775701 @default.
- W3100477570 hasOpenAccess W3100477570 @default.
- W3100477570 hasPrimaryLocation W31004775701 @default.
- W3100477570 hasRelatedWork W1713526874 @default.
- W3100477570 hasRelatedWork W1783842908 @default.
- W3100477570 hasRelatedWork W2061883802 @default.
- W3100477570 hasRelatedWork W2104854090 @default.
- W3100477570 hasRelatedWork W2105728138 @default.
- W3100477570 hasRelatedWork W2604825198 @default.
- W3100477570 hasRelatedWork W2743157634 @default.
- W3100477570 hasRelatedWork W2753258528 @default.
- W3100477570 hasRelatedWork W2774514012 @default.
- W3100477570 hasRelatedWork W2899291531 @default.
- W3100477570 hasRelatedWork W2951723130 @default.
- W3100477570 hasRelatedWork W2972385720 @default.
- W3100477570 hasRelatedWork W2981452981 @default.
- W3100477570 hasRelatedWork W3080420240 @default.
- W3100477570 hasRelatedWork W3094869948 @default.
- W3100477570 hasRelatedWork W3098902916 @default.
- W3100477570 hasRelatedWork W3127434715 @default.
- W3100477570 hasRelatedWork W3161285400 @default.