FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2947188859> ?p ?o ?g. }

• W2947188859 abstract "Recent work has uncovered the interesting (and somewhat surprising) finding that training models to be invariant to adversarial perturbations requires substantially larger datasets than those required for standard classification. This result is a key hurdle in the deployment of robust machine learning models in many real world applications where labeled data is expensive. Our main insight is that unlabeled data can be a competitive alternative to labeled data for training adversarially robust models. Theoretically, we show that in a simple statistical setting, the sample complexity for learning an adversarially robust model from unlabeled data matches the fully supervised case up to constant factors. On standard datasets like CIFAR-10, a simple Unsupervised Adversarial Training (UAT) approach using unlabeled data improves robust accuracy by 21.7% over using 4K supervised examples alone, and captures over 95% of the improvement from the same number of labeled examples. Finally, we report an improvement of 4% over the previous state-of-the-art on CIFAR-10 against the strongest known attack by using additional unlabeled data from the uncurated 80 Million Tiny Images dataset. This demonstrates that our finding extends as well to the more realistic case where unlabeled data is also uncurated, therefore opening a new avenue for improving adversarial training." @default.
• W2947188859 created "2019-06-07" @default.
• W2947188859 creator A5012826868 @default.
• W2947188859 creator A5013834379 @default.
• W2947188859 creator A5057393538 @default.
• W2947188859 creator A5059226057 @default.
• W2947188859 creator A5076040849 @default.
• W2947188859 creator A5076474156 @default.
• W2947188859 date "2019-05-31" @default.
• W2947188859 modified "2023-09-27" @default.
• W2947188859 title "Are Labels Required for Improving Adversarial Robustness" @default.
• W2947188859 cites W1479807131 @default.
• W2947188859 cites W1522301498 @default.
• W2947188859 cites W1560153690 @default.
• W2947188859 cites W1566135517 @default.
• W2947188859 cites W1883420340 @default.
• W2947188859 cites W1945616565 @default.
• W2947188859 cites W2048679005 @default.
• W2947188859 cites W2081580037 @default.
• W2947188859 cites W2129068307 @default.
• W2947188859 cites W2136189984 @default.
• W2947188859 cites W2145607950 @default.
• W2947188859 cites W2342045095 @default.
• W2947188859 cites W2342840547 @default.
• W2947188859 cites W2401231614 @default.
• W2947188859 cites W2502312327 @default.
• W2947188859 cites W2512971201 @default.
• W2947188859 cites W2618574054 @default.
• W2947188859 cites W2620038827 @default.
• W2947188859 cites W2765233338 @default.
• W2947188859 cites W2787496614 @default.
• W2947188859 cites W2787708942 @default.
• W2947188859 cites W2808139377 @default.
• W2947188859 cites W2886281300 @default.
• W2947188859 cites W2897297967 @default.
• W2947188859 cites W2898963688 @default.
• W2947188859 cites W2906989593 @default.
• W2947188859 cites W2912237282 @default.
• W2947188859 cites W2913266441 @default.
• W2947188859 cites W2924551358 @default.
• W2947188859 cites W2942203175 @default.
• W2947188859 cites W2943865428 @default.
• W2947188859 cites W2947099774 @default.
• W2947188859 cites W2947294642 @default.
• W2947188859 cites W2950864148 @default.
• W2947188859 cites W2962369866 @default.
• W2947188859 cites W2962943487 @default.
• W2947188859 cites W2962972504 @default.
• W2947188859 cites W2963289726 @default.
• W2947188859 cites W2963389226 @default.
• W2947188859 cites W2963435192 @default.
• W2947188859 cites W2963564844 @default.
• W2947188859 cites W2963777481 @default.
• W2947188859 cites W2963857521 @default.
• W2947188859 cites W2963894448 @default.
• W2947188859 cites W2963956526 @default.
• W2947188859 cites W2964153729 @default.
• W2947188859 cites W2964159205 @default.
• W2947188859 cites W2964253222 @default.
• W2947188859 cites W2964292098 @default.
• W2947188859 cites W2970680991 @default.
• W2947188859 cites W2981016037 @default.
• W2947188859 cites W3036475571 @default.
• W2947188859 cites W3118608800 @default.
• W2947188859 cites W9657784 @default.
• W2947188859 cites W2530816535 @default.
• W2947188859 hasPublicationYear "2019" @default.
• W2947188859 type Work @default.
• W2947188859 sameAs 2947188859 @default.
• W2947188859 citedByCount "27" @default.
• W2947188859 countsByYear W29471888592018 @default.
• W2947188859 countsByYear W29471888592019 @default.
• W2947188859 countsByYear W29471888592020 @default.
• W2947188859 countsByYear W29471888592021 @default.
• W2947188859 crossrefType "posted-content" @default.
• W2947188859 hasAuthorship W2947188859A5012826868 @default.
• W2947188859 hasAuthorship W2947188859A5013834379 @default.
• W2947188859 hasAuthorship W2947188859A5057393538 @default.
• W2947188859 hasAuthorship W2947188859A5059226057 @default.
• W2947188859 hasAuthorship W2947188859A5076040849 @default.
• W2947188859 hasAuthorship W2947188859A5076474156 @default.
• W2947188859 hasConcept C104317684 @default.
• W2947188859 hasConcept C105339364 @default.
• W2947188859 hasConcept C111919701 @default.
• W2947188859 hasConcept C119857082 @default.
• W2947188859 hasConcept C124101348 @default.
• W2947188859 hasConcept C153180895 @default.
• W2947188859 hasConcept C154945302 @default.
• W2947188859 hasConcept C185592680 @default.
• W2947188859 hasConcept C190470478 @default.
• W2947188859 hasConcept C2776145971 @default.
• W2947188859 hasConcept C2778445095 @default.
• W2947188859 hasConcept C33923547 @default.
• W2947188859 hasConcept C37736160 @default.
• W2947188859 hasConcept C37914503 @default.
• W2947188859 hasConcept C41008148 @default.
• W2947188859 hasConcept C51632099 @default.
• W2947188859 hasConcept C55493867 @default.
• W2947188859 hasConcept C63479239 @default.
• W2947188859 hasConceptScore W2947188859C104317684 @default.

• next