FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3015602049> ?p ?o ?g. }

• W3015602049 abstract "With the growth of interest in the attack and defense of deep neural networks, researchers are focusing more on the robustness of applying them to devices with limited memory. Thus, unlike adversarial training, which only considers the balance between accuracy and robustness, we come to a more meaningful and critical issue, i.e., the balance among accuracy, efficiency and robustness (AER). Recently, some related works focused on this issue, but with different observations, and the relations among AER remain unclear. This paper first investigates the robustness of pruned models with different compression ratios under the gradual pruning process and concludes that the robustness of the pruned model drastically varies with different pruning processes, especially in response to attacks with large strength. Second, we test the performance of mixing the clean data and adversarial examples (generated with a prescribed uniform budget) into the gradual pruning process, called adversarial pruning, and find the following: the pruned model's robustness exhibits high sensitivity to the budget. Furthermore, to better balance the AER, we propose an approach called blind adversarial pruning (BAP), which introduces the idea of blind adversarial training into the gradual pruning process. The main idea is to use a cutoff-scale strategy to adaptively estimate a nonuniform budget to modify the AEs used during pruning, thus ensuring that the strengths of AEs are dynamically located within a reasonable range at each pruning step and ultimately improving the overall AER of the pruned model. The experimental results obtained using BAP for pruning classification models based on several benchmarks demonstrate the competitive performance of this method: the robustness of the model pruned by BAP is more stable among varying pruning processes, and BAP exhibits better overall AER than adversarial pruning." @default.
• W3015602049 created "2020-04-17" @default.
• W3015602049 creator A5006257322 @default.
• W3015602049 creator A5012703091 @default.
• W3015602049 creator A5078993685 @default.
• W3015602049 creator A5089966269 @default.
• W3015602049 date "2020-04-10" @default.
• W3015602049 modified "2023-09-27" @default.
• W3015602049 title "Blind Adversarial Pruning: Balance Accuracy, Efficiency and Robustness" @default.
• W3015602049 cites W1821462560 @default.
• W3015602049 cites W1965154800 @default.
• W3015602049 cites W2035820295 @default.
• W3015602049 cites W2112796928 @default.
• W3015602049 cites W2125389748 @default.
• W3015602049 cites W2130942839 @default.
• W3015602049 cites W2134797427 @default.
• W3015602049 cites W2160815625 @default.
• W3015602049 cites W2163605009 @default.
• W3015602049 cites W2172166488 @default.
• W3015602049 cites W2183341477 @default.
• W3015602049 cites W2233116163 @default.
• W3015602049 cites W2243397390 @default.
• W3015602049 cites W2279221249 @default.
• W3015602049 cites W2300242332 @default.
• W3015602049 cites W2766839578 @default.
• W3015602049 cites W2810611310 @default.
• W3015602049 cites W2891828758 @default.
• W3015602049 cites W2897355816 @default.
• W3015602049 cites W2899748887 @default.
• W3015602049 cites W2900959181 @default.
• W3015602049 cites W2907518016 @default.
• W3015602049 cites W2962700793 @default.
• W3015602049 cites W2962761044 @default.
• W3015602049 cites W2963114950 @default.
• W3015602049 cites W2963207607 @default.
• W3015602049 cites W2963389226 @default.
• W3015602049 cites W2963597678 @default.
• W3015602049 cites W2963674932 @default.
• W3015602049 cites W2963744840 @default.
• W3015602049 cites W2963752115 @default.
• W3015602049 cites W2963770044 @default.
• W3015602049 cites W2964014389 @default.
• W3015602049 cites W2964118293 @default.
• W3015602049 cites W2964121744 @default.
• W3015602049 cites W2964153729 @default.
• W3015602049 cites W2964224652 @default.
• W3015602049 cites W2964253222 @default.
• W3015602049 cites W2964299589 @default.
• W3015602049 cites W2970330753 @default.
• W3015602049 cites W2971229607 @default.
• W3015602049 cites W3002772968 @default.
• W3015602049 cites W3009751875 @default.
• W3015602049 cites W3101114581 @default.
• W3015602049 cites W3118608800 @default.
• W3015602049 cites W3146803896 @default.
• W3015602049 hasPublicationYear "2020" @default.
• W3015602049 type Work @default.
• W3015602049 sameAs 3015602049 @default.
• W3015602049 citedByCount "0" @default.
• W3015602049 crossrefType "proceedings-article" @default.
• W3015602049 hasAuthorship W3015602049A5006257322 @default.
• W3015602049 hasAuthorship W3015602049A5012703091 @default.
• W3015602049 hasAuthorship W3015602049A5078993685 @default.
• W3015602049 hasAuthorship W3015602049A5089966269 @default.
• W3015602049 hasBestOaLocation W30156020491 @default.
• W3015602049 hasConcept C104317684 @default.
• W3015602049 hasConcept C108010975 @default.
• W3015602049 hasConcept C119857082 @default.
• W3015602049 hasConcept C154945302 @default.
• W3015602049 hasConcept C185592680 @default.
• W3015602049 hasConcept C37736160 @default.
• W3015602049 hasConcept C41008148 @default.
• W3015602049 hasConcept C50644808 @default.
• W3015602049 hasConcept C55493867 @default.
• W3015602049 hasConcept C63479239 @default.
• W3015602049 hasConcept C6557445 @default.
• W3015602049 hasConcept C86803240 @default.
• W3015602049 hasConceptScore W3015602049C104317684 @default.
• W3015602049 hasConceptScore W3015602049C108010975 @default.
• W3015602049 hasConceptScore W3015602049C119857082 @default.
• W3015602049 hasConceptScore W3015602049C154945302 @default.
• W3015602049 hasConceptScore W3015602049C185592680 @default.
• W3015602049 hasConceptScore W3015602049C37736160 @default.
• W3015602049 hasConceptScore W3015602049C41008148 @default.
• W3015602049 hasConceptScore W3015602049C50644808 @default.
• W3015602049 hasConceptScore W3015602049C55493867 @default.
• W3015602049 hasConceptScore W3015602049C63479239 @default.
• W3015602049 hasConceptScore W3015602049C6557445 @default.
• W3015602049 hasConceptScore W3015602049C86803240 @default.
• W3015602049 hasLocation W30156020491 @default.
• W3015602049 hasOpenAccess W3015602049 @default.
• W3015602049 hasPrimaryLocation W30156020491 @default.
• W3015602049 hasRelatedWork W2903329274 @default.
• W3015602049 hasRelatedWork W2924014439 @default.
• W3015602049 hasRelatedWork W2950630935 @default.
• W3015602049 hasRelatedWork W2953355478 @default.
• W3015602049 hasRelatedWork W2964540166 @default.
• W3015602049 hasRelatedWork W2979999556 @default.
• W3015602049 hasRelatedWork W2982561504 @default.
• W3015602049 hasRelatedWork W3007731245 @default.

• next