FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386065877> ?p ?o ?g. }

• W4386065877 abstract "Recent studies have shown that higher accuracy on ImageNet usually leads to better robustness against different corruptions. Therefore, in this paper, instead of following the traditional research paradigm that investigates new out-of-distribution corruptions or perturbations deep models may encounter, we conduct model debugging in in-distribution data to explore which object attributes a model may be sensitive to. To achieve this goal, we create a toolkit for object editing with controls of backgrounds, sizes, positions, and directions, and create a rigorous benchmark named ImageNet-E(diting) for evaluating the image classifier robustness in terms of object attributes. With our ImageNet-E, we evaluate the performance of current deep learning models, including both convolutional neural networks and vision transformers. We find that most models are quite sensitive to attribute changes. A small change in the background can lead to an average of 9.23% drop on top-1 accuracy. We also evaluate some robust models including both adversarially trained models and other robust trained models and find that some models show worse robustness against attribute changes than vanilla models. Based on these findings, we discover ways to enhance attribute robustness with preprocessing, architecture designs, and training strategies. We hope this work can provide some insights to the community and open up a new avenue for research in robust computer vision. The code and dataset are available at https://github.com/alibaba/easyrobust." @default.
• W4386065877 created "2023-08-23" @default.
• W4386065877 creator A5034112897 @default.
• W4386065877 creator A5034764096 @default.
• W4386065877 creator A5042884230 @default.
• W4386065877 creator A5071690624 @default.
• W4386065877 creator A5071894027 @default.
• W4386065877 creator A5083967189 @default.
• W4386065877 date "2023-06-01" @default.
• W4386065877 modified "2023-10-18" @default.
• W4386065877 title "ImageNet-E: Benchmarking Neural Network Robustness via Attribute Editing" @default.
• W4386065877 cites W2044465660 @default.
• W4386065877 cites W2048971218 @default.
• W4386065877 cites W2117539524 @default.
• W4386065877 cites W2194775991 @default.
• W4386065877 cites W2592929672 @default.
• W4386065877 cites W2963446712 @default.
• W4386065877 cites W2963564844 @default.
• W4386065877 cites W3016719260 @default.
• W4386065877 cites W3016970897 @default.
• W4386065877 cites W3022746105 @default.
• W4386065877 cites W3035160371 @default.
• W4386065877 cites W3036286896 @default.
• W4386065877 cites W3037492894 @default.
• W4386065877 cites W3116074996 @default.
• W4386065877 cites W3138516171 @default.
• W4386065877 cites W3145450063 @default.
• W4386065877 cites W3176482836 @default.
• W4386065877 cites W3191453585 @default.
• W4386065877 cites W3198675127 @default.
• W4386065877 cites W3216352822 @default.
• W4386065877 cites W4213447687 @default.
• W4386065877 cites W4304701424 @default.
• W4386065877 cites W4312443924 @default.
• W4386065877 cites W4312705932 @default.
• W4386065877 cites W4312771828 @default.
• W4386065877 cites W4312804044 @default.
• W4386065877 cites W4313156423 @default.
• W4386065877 doi "https://doi.org/10.1109/cvpr52729.2023.01951" @default.
• W4386065877 hasPublicationYear "2023" @default.
• W4386065877 type Work @default.
• W4386065877 citedByCount "0" @default.
• W4386065877 crossrefType "proceedings-article" @default.
• W4386065877 hasAuthorship W4386065877A5034112897 @default.
• W4386065877 hasAuthorship W4386065877A5034764096 @default.
• W4386065877 hasAuthorship W4386065877A5042884230 @default.
• W4386065877 hasAuthorship W4386065877A5071690624 @default.
• W4386065877 hasAuthorship W4386065877A5071894027 @default.
• W4386065877 hasAuthorship W4386065877A5083967189 @default.
• W4386065877 hasConcept C104317684 @default.
• W4386065877 hasConcept C108583219 @default.
• W4386065877 hasConcept C111919701 @default.
• W4386065877 hasConcept C119857082 @default.
• W4386065877 hasConcept C123657996 @default.
• W4386065877 hasConcept C124101348 @default.
• W4386065877 hasConcept C142362112 @default.
• W4386065877 hasConcept C144133560 @default.
• W4386065877 hasConcept C153349607 @default.
• W4386065877 hasConcept C154945302 @default.
• W4386065877 hasConcept C162853370 @default.
• W4386065877 hasConcept C168065819 @default.
• W4386065877 hasConcept C185592680 @default.
• W4386065877 hasConcept C199360897 @default.
• W4386065877 hasConcept C34736171 @default.
• W4386065877 hasConcept C41008148 @default.
• W4386065877 hasConcept C43126263 @default.
• W4386065877 hasConcept C50644808 @default.
• W4386065877 hasConcept C55493867 @default.
• W4386065877 hasConcept C63479239 @default.
• W4386065877 hasConcept C81363708 @default.
• W4386065877 hasConcept C86251818 @default.
• W4386065877 hasConceptScore W4386065877C104317684 @default.
• W4386065877 hasConceptScore W4386065877C108583219 @default.
• W4386065877 hasConceptScore W4386065877C111919701 @default.
• W4386065877 hasConceptScore W4386065877C119857082 @default.
• W4386065877 hasConceptScore W4386065877C123657996 @default.
• W4386065877 hasConceptScore W4386065877C124101348 @default.
• W4386065877 hasConceptScore W4386065877C142362112 @default.
• W4386065877 hasConceptScore W4386065877C144133560 @default.
• W4386065877 hasConceptScore W4386065877C153349607 @default.
• W4386065877 hasConceptScore W4386065877C154945302 @default.
• W4386065877 hasConceptScore W4386065877C162853370 @default.
• W4386065877 hasConceptScore W4386065877C168065819 @default.
• W4386065877 hasConceptScore W4386065877C185592680 @default.
• W4386065877 hasConceptScore W4386065877C199360897 @default.
• W4386065877 hasConceptScore W4386065877C34736171 @default.
• W4386065877 hasConceptScore W4386065877C41008148 @default.
• W4386065877 hasConceptScore W4386065877C43126263 @default.
• W4386065877 hasConceptScore W4386065877C50644808 @default.
• W4386065877 hasConceptScore W4386065877C55493867 @default.
• W4386065877 hasConceptScore W4386065877C63479239 @default.
• W4386065877 hasConceptScore W4386065877C81363708 @default.
• W4386065877 hasConceptScore W4386065877C86251818 @default.
• W4386065877 hasFunder F4320335777 @default.
• W4386065877 hasLocation W43860658771 @default.
• W4386065877 hasOpenAccess W4386065877 @default.
• W4386065877 hasPrimaryLocation W43860658771 @default.
• W4386065877 hasRelatedWork W2731899572 @default.
• W4386065877 hasRelatedWork W2999805992 @default.
• W4386065877 hasRelatedWork W3116150086 @default.

• next