FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2947433239> ?p ?o ?g. }

• W2947433239 abstract "Ensemble approaches for uncertainty estimation have recently been applied to the tasks of misclassification detection, out-of-distribution input detection and adversarial attack detection. Prior Networks have been proposed as an approach to efficiently emph{emulate} an ensemble of models for classification by parameterising a Dirichlet prior distribution over output distributions. These models have been shown to outperform alternative ensemble approaches, such as Monte-Carlo Dropout, on the task of out-of-distribution input detection. However, scaling Prior Networks to complex datasets with many classes is difficult using the training criteria originally proposed. This paper makes two contributions. First, we show that the appropriate training criterion for Prior Networks is the emph{reverse} KL-divergence between Dirichlet distributions. This addresses issues in the nature of the training data target distributions, enabling prior networks to be successfully trained on classification tasks with arbitrarily many classes, as well as improving out-of-distribution detection performance. Second, taking advantage of this new training criterion, this paper investigates using Prior Networks to detect adversarial attacks and proposes a generalized form of adversarial training. It is shown that the construction of successful emph{adaptive} whitebox attacks, which affect the prediction and evade detection, against Prior Networks trained on CIFAR-10 and CIFAR-100 using the proposed approach requires a greater amount of computational effort than against networks defended using standard adversarial training or MC-dropout." @default.
• W2947433239 created "2019-06-07" @default.
• W2947433239 creator A5050766679 @default.
• W2947433239 creator A5080001860 @default.
• W2947433239 date "2019-05-31" @default.
• W2947433239 modified "2023-09-27" @default.
• W2947433239 title "Reverse KL-Divergence Training of Prior Networks: Improved Uncertainty and Adversarial Robustness" @default.
• W2947433239 cites W1019830208 @default.
• W2947433239 cites W1536680647 @default.
• W2947433239 cites W179875071 @default.
• W2947433239 cites W1922655562 @default.
• W2947433239 cites W1959608418 @default.
• W2947433239 cites W1996796871 @default.
• W2947433239 cites W2130306094 @default.
• W2947433239 cites W2141599568 @default.
• W2947433239 cites W2180612164 @default.
• W2947433239 cites W2184045248 @default.
• W2947433239 cites W2274565976 @default.
• W2947433239 cites W2516574342 @default.
• W2947433239 cites W2570685808 @default.
• W2947433239 cites W2590523583 @default.
• W2947433239 cites W2605631833 @default.
• W2947433239 cites W2640329709 @default.
• W2947433239 cites W2765867474 @default.
• W2947433239 cites W2774644650 @default.
• W2947433239 cites W2806471870 @default.
• W2947433239 cites W2913848079 @default.
• W2947433239 cites W2950468330 @default.
• W2947433239 cites W2950577311 @default.
• W2947433239 cites W2952053192 @default.
• W2947433239 cites W2962835968 @default.
• W2947433239 cites W2963207607 @default.
• W2947433239 cites W2963215553 @default.
• W2947433239 cites W2963238274 @default.
• W2947433239 cites W2963253230 @default.
• W2947433239 cites W2963542245 @default.
• W2947433239 cites W2963564844 @default.
• W2947433239 cites W2963612069 @default.
• W2947433239 cites W2964059111 @default.
• W2947433239 cites W2964082701 @default.
• W2947433239 cites W2964121744 @default.
• W2947433239 cites W2964317943 @default.
• W2947433239 cites W2996184071 @default.
• W2947433239 cites W3118608800 @default.
• W2947433239 hasPublicationYear "2019" @default.
• W2947433239 type Work @default.
• W2947433239 sameAs 2947433239 @default.
• W2947433239 citedByCount "11" @default.
• W2947433239 countsByYear W29474332392019 @default.
• W2947433239 countsByYear W29474332392020 @default.
• W2947433239 countsByYear W29474332392021 @default.
• W2947433239 crossrefType "posted-content" @default.
• W2947433239 hasAuthorship W2947433239A5050766679 @default.
• W2947433239 hasAuthorship W2947433239A5080001860 @default.
• W2947433239 hasConcept C104317684 @default.
• W2947433239 hasConcept C105795698 @default.
• W2947433239 hasConcept C119857082 @default.
• W2947433239 hasConcept C121332964 @default.
• W2947433239 hasConcept C134306372 @default.
• W2947433239 hasConcept C138885662 @default.
• W2947433239 hasConcept C153294291 @default.
• W2947433239 hasConcept C154945302 @default.
• W2947433239 hasConcept C169214877 @default.
• W2947433239 hasConcept C182310444 @default.
• W2947433239 hasConcept C185592680 @default.
• W2947433239 hasConcept C19499675 @default.
• W2947433239 hasConcept C207390915 @default.
• W2947433239 hasConcept C2776145597 @default.
• W2947433239 hasConcept C2777211547 @default.
• W2947433239 hasConcept C33923547 @default.
• W2947433239 hasConcept C37736160 @default.
• W2947433239 hasConcept C41008148 @default.
• W2947433239 hasConcept C41895202 @default.
• W2947433239 hasConcept C55493867 @default.
• W2947433239 hasConcept C63479239 @default.
• W2947433239 hasConceptScore W2947433239C104317684 @default.
• W2947433239 hasConceptScore W2947433239C105795698 @default.
• W2947433239 hasConceptScore W2947433239C119857082 @default.
• W2947433239 hasConceptScore W2947433239C121332964 @default.
• W2947433239 hasConceptScore W2947433239C134306372 @default.
• W2947433239 hasConceptScore W2947433239C138885662 @default.
• W2947433239 hasConceptScore W2947433239C153294291 @default.
• W2947433239 hasConceptScore W2947433239C154945302 @default.
• W2947433239 hasConceptScore W2947433239C169214877 @default.
• W2947433239 hasConceptScore W2947433239C182310444 @default.
• W2947433239 hasConceptScore W2947433239C185592680 @default.
• W2947433239 hasConceptScore W2947433239C19499675 @default.
• W2947433239 hasConceptScore W2947433239C207390915 @default.
• W2947433239 hasConceptScore W2947433239C2776145597 @default.
• W2947433239 hasConceptScore W2947433239C2777211547 @default.
• W2947433239 hasConceptScore W2947433239C33923547 @default.
• W2947433239 hasConceptScore W2947433239C37736160 @default.
• W2947433239 hasConceptScore W2947433239C41008148 @default.
• W2947433239 hasConceptScore W2947433239C41895202 @default.
• W2947433239 hasConceptScore W2947433239C55493867 @default.
• W2947433239 hasConceptScore W2947433239C63479239 @default.
• W2947433239 hasLocation W29474332391 @default.
• W2947433239 hasOpenAccess W2947433239 @default.
• W2947433239 hasPrimaryLocation W29474332391 @default.
• W2947433239 hasRelatedWork W2269778407 @default.

• next