FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3013068160> ?p ?o ?g. }

• W3013068160 abstract "Machine learning models are prone to memorizing sensitive data, making them vulnerable to membership inference attacks in which an adversary aims to guess if an input sample was used to train the model. In this paper, we show that prior work on membership inference attacks may severely underestimate the privacy risks by relying solely on training custom neural network classifiers to perform attacks and focusing only on the aggregate results over data samples, such as the attack accuracy. To overcome these limitations, we first propose to benchmark membership inference privacy risks by improving existing non-neural network based inference attacks and proposing a new inference attack method based on a modification of prediction entropy. We also propose benchmarks for defense mechanisms by accounting for adaptive adversaries with knowledge of the defense and also accounting for the trade-off between model accuracy and privacy risks. Using our benchmark attacks, we demonstrate that existing defense approaches are not as effective as previously reported. Next, we introduce a new approach for fine-grained privacy analysis by formulating and deriving a new metric called the privacy risk score. Our privacy risk score metric measures an individual sample's likelihood of being a training member, which allows an adversary to identify samples with high privacy risks and perform attacks with high confidence. We experimentally validate the effectiveness of the privacy risk score and demonstrate that the distribution of privacy risk score across individual samples is heterogeneous. Finally, we perform an in-depth investigation for understanding why certain samples have high privacy risks, including correlations with model sensitivity, generalization error, and feature embeddings. Our work emphasizes the importance of a systematic and rigorous evaluation of privacy risks of machine learning models." @default.
• W3013068160 created "2020-04-03" @default.
• W3013068160 creator A5015619835 @default.
• W3013068160 creator A5074489729 @default.
• W3013068160 date "2020-03-24" @default.
• W3013068160 modified "2023-09-27" @default.
• W3013068160 title "Systematic Evaluation of Privacy Risks of Machine Learning Models" @default.
• W3013068160 cites W1582774210 @default.
• W3013068160 cites W1873763122 @default.
• W3013068160 cites W2034978228 @default.
• W3013068160 cites W2051267297 @default.
• W3013068160 cites W2053637704 @default.
• W3013068160 cites W2095705004 @default.
• W3013068160 cites W2108598243 @default.
• W3013068160 cites W2144513243 @default.
• W3013068160 cites W2156876426 @default.
• W3013068160 cites W2187089797 @default.
• W3013068160 cites W2194775991 @default.
• W3013068160 cites W2473418344 @default.
• W3013068160 cites W2532781556 @default.
• W3013068160 cites W2535690855 @default.
• W3013068160 cites W2591882872 @default.
• W3013068160 cites W2757528734 @default.
• W3013068160 cites W2784621220 @default.
• W3013068160 cites W2786233556 @default.
• W3013068160 cites W2795435272 @default.
• W3013068160 cites W2798657499 @default.
• W3013068160 cites W2884943453 @default.
• W3013068160 cites W2887995258 @default.
• W3013068160 cites W2897830718 @default.
• W3013068160 cites W2946930197 @default.
• W3013068160 cites W2952270003 @default.
• W3013068160 cites W2962835968 @default.
• W3013068160 cites W2963143631 @default.
• W3013068160 cites W2963378725 @default.
• W3013068160 cites W2963431851 @default.
• W3013068160 cites W2963446712 @default.
• W3013068160 cites W2963456518 @default.
• W3013068160 cites W2963543276 @default.
• W3013068160 cites W2963564844 @default.
• W3013068160 cites W2964151798 @default.
• W3013068160 cites W2964253222 @default.
• W3013068160 cites W2965527189 @default.
• W3013068160 cites W2967985550 @default.
• W3013068160 cites W2970024769 @default.
• W3013068160 cites W2972280210 @default.
• W3013068160 cites W2983140679 @default.
• W3013068160 cites W3048684575 @default.
• W3013068160 cites W3048775464 @default.
• W3013068160 cites W3104224589 @default.
• W3013068160 cites W3118608800 @default.
• W3013068160 hasPublicationYear "2020" @default.
• W3013068160 type Work @default.
• W3013068160 sameAs 3013068160 @default.
• W3013068160 citedByCount "3" @default.
• W3013068160 countsByYear W30130681602020 @default.
• W3013068160 countsByYear W30130681602021 @default.
• W3013068160 crossrefType "posted-content" @default.
• W3013068160 hasAuthorship W3013068160A5015619835 @default.
• W3013068160 hasAuthorship W3013068160A5074489729 @default.
• W3013068160 hasConcept C106301342 @default.
• W3013068160 hasConcept C119857082 @default.
• W3013068160 hasConcept C121332964 @default.
• W3013068160 hasConcept C124101348 @default.
• W3013068160 hasConcept C127413603 @default.
• W3013068160 hasConcept C13280743 @default.
• W3013068160 hasConcept C154945302 @default.
• W3013068160 hasConcept C176217482 @default.
• W3013068160 hasConcept C185592680 @default.
• W3013068160 hasConcept C185798385 @default.
• W3013068160 hasConcept C198531522 @default.
• W3013068160 hasConcept C205649164 @default.
• W3013068160 hasConcept C21547014 @default.
• W3013068160 hasConcept C2776214188 @default.
• W3013068160 hasConcept C38652104 @default.
• W3013068160 hasConcept C41008148 @default.
• W3013068160 hasConcept C41065033 @default.
• W3013068160 hasConcept C43617362 @default.
• W3013068160 hasConcept C50644808 @default.
• W3013068160 hasConcept C62520636 @default.
• W3013068160 hasConceptScore W3013068160C106301342 @default.
• W3013068160 hasConceptScore W3013068160C119857082 @default.
• W3013068160 hasConceptScore W3013068160C121332964 @default.
• W3013068160 hasConceptScore W3013068160C124101348 @default.
• W3013068160 hasConceptScore W3013068160C127413603 @default.
• W3013068160 hasConceptScore W3013068160C13280743 @default.
• W3013068160 hasConceptScore W3013068160C154945302 @default.
• W3013068160 hasConceptScore W3013068160C176217482 @default.
• W3013068160 hasConceptScore W3013068160C185592680 @default.
• W3013068160 hasConceptScore W3013068160C185798385 @default.
• W3013068160 hasConceptScore W3013068160C198531522 @default.
• W3013068160 hasConceptScore W3013068160C205649164 @default.
• W3013068160 hasConceptScore W3013068160C21547014 @default.
• W3013068160 hasConceptScore W3013068160C2776214188 @default.
• W3013068160 hasConceptScore W3013068160C38652104 @default.
• W3013068160 hasConceptScore W3013068160C41008148 @default.
• W3013068160 hasConceptScore W3013068160C41065033 @default.
• W3013068160 hasConceptScore W3013068160C43617362 @default.
• W3013068160 hasConceptScore W3013068160C50644808 @default.
• W3013068160 hasConceptScore W3013068160C62520636 @default.

• next