Matches in SemOpenAlex for { <https://semopenalex.org/work/W3129409966> ?p ?o ?g. }
- W3129409966 abstract "When receiving machine learning services from the cloud, the provider does not need to receive all features; in fact, only a subset of the features are necessary for the target prediction task. Discerning this subset is the key problem of this work. We formulate this problem as a gradient-based perturbation maximization method that discovers this subset in the input feature space with respect to the functionality of the prediction model used by the provider. After identifying the subset, our framework, Cloak, suppresses the rest of the features using utility-preserving constant values that are discovered through a separate gradient-based optimization process. We show that Cloak does not necessarily require collaboration from the service provider beyond its normal service, and can be applied in scenarios where we only have black-box access to the service provider's model. We theoretically guarantee that Cloak's optimizations reduce the upper bound of the Mutual Information (MI) between the data and the sifted representations that are sent out. Experimental results show that Cloak reduces the mutual information between the input and the sifted representations by 85.01% with only a negligible reduction in utility (1.42%). In addition, we show that Cloak greatly diminishes adversaries' ability to learn and infer non-conducive features." @default.
- W3129409966 created "2021-03-01" @default.
- W3129409966 creator A5014947998 @default.
- W3129409966 creator A5018043043 @default.
- W3129409966 creator A5030613735 @default.
- W3129409966 creator A5055319560 @default.
- W3129409966 creator A5065583627 @default.
- W3129409966 creator A5084514143 @default.
- W3129409966 date "2020-03-26" @default.
- W3129409966 modified "2023-09-27" @default.
- W3129409966 title "Not All Features Are Equal: Discovering Essential Features for Preserving Prediction Privacy." @default.
- W3129409966 cites W1522345630 @default.
- W3129409966 cites W1557833142 @default.
- W3129409966 cites W1579792977 @default.
- W3129409966 cites W1686810756 @default.
- W3129409966 cites W1770244604 @default.
- W3129409966 cites W1834627138 @default.
- W3129409966 cites W1873763122 @default.
- W3129409966 cites W2027595342 @default.
- W3129409966 cites W2031533839 @default.
- W3129409966 cites W2053637704 @default.
- W3129409966 cites W2099111195 @default.
- W3129409966 cites W2100773064 @default.
- W3129409966 cites W2103133870 @default.
- W3129409966 cites W2112380340 @default.
- W3129409966 cites W2112796928 @default.
- W3129409966 cites W2144190046 @default.
- W3129409966 cites W2194775991 @default.
- W3129409966 cites W2233194383 @default.
- W3129409966 cites W2416572683 @default.
- W3129409966 cites W2435473771 @default.
- W3129409966 cites W2473418344 @default.
- W3129409966 cites W2535690855 @default.
- W3129409966 cites W2545564314 @default.
- W3129409966 cites W2592232824 @default.
- W3129409966 cites W2593324641 @default.
- W3129409966 cites W2611576673 @default.
- W3129409966 cites W2623464795 @default.
- W3129409966 cites W2701059868 @default.
- W3129409966 cites W2742675945 @default.
- W3129409966 cites W2765200655 @default.
- W3129409966 cites W2767079719 @default.
- W3129409966 cites W2785361959 @default.
- W3129409966 cites W2788416960 @default.
- W3129409966 cites W2792106437 @default.
- W3129409966 cites W2805074088 @default.
- W3129409966 cites W2806377849 @default.
- W3129409966 cites W2883613460 @default.
- W3129409966 cites W2884827599 @default.
- W3129409966 cites W2886576854 @default.
- W3129409966 cites W2913257450 @default.
- W3129409966 cites W2917560727 @default.
- W3129409966 cites W2950306824 @default.
- W3129409966 cites W2950602864 @default.
- W3129409966 cites W2950943617 @default.
- W3129409966 cites W2951266961 @default.
- W3129409966 cites W2960248881 @default.
- W3129409966 cites W2963096987 @default.
- W3129409966 cites W2963311060 @default.
- W3129409966 cites W2963733194 @default.
- W3129409966 cites W2963963190 @default.
- W3129409966 cites W2964121744 @default.
- W3129409966 cites W2970880934 @default.
- W3129409966 cites W2974492900 @default.
- W3129409966 cites W2974546631 @default.
- W3129409966 cites W2981828710 @default.
- W3129409966 cites W2983140679 @default.
- W3129409966 cites W2984644945 @default.
- W3129409966 cites W2985109827 @default.
- W3129409966 cites W2989885118 @default.
- W3129409966 cites W2995022099 @default.
- W3129409966 cites W2998600867 @default.
- W3129409966 cites W3011434423 @default.
- W3129409966 cites W3012949032 @default.
- W3129409966 cites W3013099760 @default.
- W3129409966 cites W3014643566 @default.
- W3129409966 cites W3018763859 @default.
- W3129409966 cites W3021878953 @default.
- W3129409966 cites W3033488952 @default.
- W3129409966 cites W3033700072 @default.
- W3129409966 cites W3034162411 @default.
- W3129409966 cites W3035493623 @default.
- W3129409966 cites W3042422567 @default.
- W3129409966 cites W3043571714 @default.
- W3129409966 cites W3080822276 @default.
- W3129409966 cites W3081488247 @default.
- W3129409966 cites W3087503988 @default.
- W3129409966 cites W3088268709 @default.
- W3129409966 cites W3094488859 @default.
- W3129409966 cites W3099785009 @default.
- W3129409966 cites W3102303514 @default.
- W3129409966 cites W3118608800 @default.
- W3129409966 cites W3217629331 @default.
- W3129409966 cites W3145167355 @default.
- W3129409966 hasPublicationYear "2020" @default.
- W3129409966 type Work @default.
- W3129409966 sameAs 3129409966 @default.
- W3129409966 citedByCount "6" @default.
- W3129409966 countsByYear W31294099662020 @default.
- W3129409966 countsByYear W31294099662021 @default.