FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3213231502> ?p ?o ?g. }

• W3213231502 abstract "Deep neural networks are susceptible to adversarially crafted, small and imperceptible changes in the natural inputs. The most effective defense mechanism against these examples is adversarial training which constructs adversarial examples during training by iterative maximization of loss. The model is then trained to minimize the loss on these constructed examples. This min-max optimization requires more data, larger capacity models, and additional computing resources. It also degrades the standard generalization performance of a model. Can we achieve robustness more efficiently? In this work, we explore this question from the perspective of knowledge transfer. First, we theoretically show the transferability of robustness from an adversarially trained teacher model to a student model with the help of mixup augmentation. Second, we propose a novel robustness transfer method called Mixup-Based Activated Channel Maps (MixACM) Transfer. MixACM transfers robustness from a robust teacher to a student by matching activated channel maps generated without expensive adversarial perturbations. Finally, extensive experiments on multiple datasets and different learning scenarios show our method can transfer robustness while also improving generalization on natural images." @default.
• W3213231502 created "2021-11-22" @default.
• W3213231502 creator A5036381691 @default.
• W3213231502 creator A5036948731 @default.
• W3213231502 creator A5052677272 @default.
• W3213231502 creator A5065103755 @default.
• W3213231502 creator A5077862962 @default.
• W3213231502 creator A5082500328 @default.
• W3213231502 date "2021-11-09" @default.
• W3213231502 modified "2023-09-22" @default.
• W3213231502 title "MixACM: Mixup-Based Robustness Transfer via Distillation of Activated Channel Maps" @default.
• W3213231502 cites W1821462560 @default.
• W3213231502 cites W1883420340 @default.
• W3213231502 cites W2108598243 @default.
• W3213231502 cites W2194775991 @default.
• W3213231502 cites W2460937040 @default.
• W3213231502 cites W2603766943 @default.
• W3213231502 cites W2606101940 @default.
• W3213231502 cites W2731516819 @default.
• W3213231502 cites W2739879705 @default.
• W3213231502 cites W2765384636 @default.
• W3213231502 cites W2765407302 @default.
• W3213231502 cites W2774018344 @default.
• W3213231502 cites W2786118190 @default.
• W3213231502 cites W2884491628 @default.
• W3213231502 cites W2897355816 @default.
• W3213231502 cites W2907518016 @default.
• W3213231502 cites W2912237282 @default.
• W3213231502 cites W2934684307 @default.
• W3213231502 cites W2944870588 @default.
• W3213231502 cites W2945359720 @default.
• W3213231502 cites W2946089113 @default.
• W3213231502 cites W2946376184 @default.
• W3213231502 cites W2946498182 @default.
• W3213231502 cites W2947294642 @default.
• W3213231502 cites W2947775933 @default.
• W3213231502 cites W2949829435 @default.
• W3213231502 cites W2952322298 @default.
• W3213231502 cites W2962729158 @default.
• W3213231502 cites W2962872506 @default.
• W3213231502 cites W2962972504 @default.
• W3213231502 cites W2963143631 @default.
• W3213231502 cites W2963207607 @default.
• W3213231502 cites W2963249138 @default.
• W3213231502 cites W2963263347 @default.
• W3213231502 cites W2963397674 @default.
• W3213231502 cites W2963744840 @default.
• W3213231502 cites W2963857521 @default.
• W3213231502 cites W2963921882 @default.
• W3213231502 cites W2964082701 @default.
• W3213231502 cites W2964111476 @default.
• W3213231502 cites W2964118293 @default.
• W3213231502 cites W2964153729 @default.
• W3213231502 cites W2964224652 @default.
• W3213231502 cites W2964253222 @default.
• W3213231502 cites W2964276371 @default.
• W3213231502 cites W2970088379 @default.
• W3213231502 cites W2970115835 @default.
• W3213231502 cites W2970317235 @default.
• W3213231502 cites W2970680991 @default.
• W3213231502 cites W2971316968 @default.
• W3213231502 cites W2976501124 @default.
• W3213231502 cites W2982157312 @default.
• W3213231502 cites W2982242214 @default.
• W3213231502 cites W2995056242 @default.
• W3213231502 cites W2995319142 @default.
• W3213231502 cites W2995607862 @default.
• W3213231502 cites W2996325578 @default.
• W3213231502 cites W2996344901 @default.
• W3213231502 cites W2996568780 @default.
• W3213231502 cites W2998835636 @default.
• W3213231502 cites W3009153202 @default.
• W3213231502 cites W3009751875 @default.
• W3213231502 cites W3021182036 @default.
• W3213231502 cites W3034444660 @default.
• W3213231502 cites W3034537217 @default.
• W3213231502 cites W3034758058 @default.
• W3213231502 cites W3034994123 @default.
• W3213231502 cites W3035032188 @default.
• W3213231502 cites W3035345420 @default.
• W3213231502 cites W3035590812 @default.
• W3213231502 cites W3035682985 @default.
• W3213231502 cites W3037742886 @default.
• W3213231502 cites W3082665562 @default.
• W3213231502 cites W3090285675 @default.
• W3213231502 cites W3092171228 @default.
• W3213231502 cites W3098018961 @default.
• W3213231502 cites W3098628719 @default.
• W3213231502 cites W3098632754 @default.
• W3213231502 cites W3099502074 @default.
• W3213231502 cites W3100995366 @default.
• W3213231502 cites W3101114581 @default.
• W3213231502 cites W3103385169 @default.
• W3213231502 cites W3104620392 @default.
• W3213231502 cites W3107235539 @default.
• W3213231502 cites W3122730565 @default.
• W3213231502 cites W3123856301 @default.
• W3213231502 cites W3125614726 @default.
• W3213231502 cites W3125645205 @default.
• W3213231502 cites W3131606849 @default.

• next