Matches in SemOpenAlex for { <https://semopenalex.org/work/W3215093302> ?p ?o ?g. }
Showing items 1 to 86 of
86
with 100 items per page.
- W3215093302 endingPage "1958" @default.
- W3215093302 startingPage "1925" @default.
- W3215093302 abstract "The embedding and extraction of useful knowledge is a recent trend in machine learning applications, e.g., to supplement existing datasets that are small. Whilst, as the increasing use of machine learning models in security-critical applications, the embedding and extraction of malicious knowledge are equivalent to the notorious backdoor attack and its defence, respectively. This paper studies the embedding and extraction of knowledge in tree ensemble classifiers, and focuses on knowledge expressible with a generic form of Boolean formulas, e.g., robustness properties and backdoor attacks. For the embedding, it is required to be preservative(the original performance of the classifier is preserved), verifiable(the knowledge can be attested), and stealthy(the embedding cannot be easily detected). To facilitate this, we propose two novel, and effective, embedding algorithms, one of which is for black-box settings and the other for white-box settings.The embedding can be done in PTIME. Beyond the embedding, we develop an algorithm to extract the embedded knowledge, by reducing the problem to be solvable with an SMT (satisfiability modulo theories) solver. While this novel algorithm can successfully extract knowledge, the reduction leads to an NP computation. Therefore, if applying embedding as backdoor attacks and extraction as defence, our results suggest a complexity gap (P vs. NP) between the attack and defence when working with tree ensemble classifiers. We apply our algorithms toa diverse set of datasets to validate our conclusion extensively." @default.
- W3215093302 created "2021-12-06" @default.
- W3215093302 creator A5020085889 @default.
- W3215093302 creator A5033093547 @default.
- W3215093302 creator A5090309356 @default.
- W3215093302 date "2021-11-24" @default.
- W3215093302 modified "2023-10-16" @default.
- W3215093302 title "Embedding and extraction of knowledge in tree ensemble classifiers" @default.
- W3215093302 cites W1748932423 @default.
- W3215093302 cites W2060375854 @default.
- W3215093302 cites W2113242816 @default.
- W3215093302 cites W2153635508 @default.
- W3215093302 cites W2167370496 @default.
- W3215093302 cites W2604242010 @default.
- W3215093302 cites W2753783305 @default.
- W3215093302 cites W2788730650 @default.
- W3215093302 cites W2803414046 @default.
- W3215093302 cites W2807363941 @default.
- W3215093302 cites W2942091739 @default.
- W3215093302 cites W2942347321 @default.
- W3215093302 cites W2957144537 @default.
- W3215093302 cites W2963793947 @default.
- W3215093302 cites W2971223760 @default.
- W3215093302 cites W2974383344 @default.
- W3215093302 cites W2987678574 @default.
- W3215093302 cites W2990347604 @default.
- W3215093302 cites W2997425368 @default.
- W3215093302 cites W3012300836 @default.
- W3215093302 cites W3015815227 @default.
- W3215093302 cites W3037471945 @default.
- W3215093302 cites W4252979261 @default.
- W3215093302 doi "https://doi.org/10.1007/s10994-021-06068-6" @default.
- W3215093302 hasPublicationYear "2021" @default.
- W3215093302 type Work @default.
- W3215093302 sameAs 3215093302 @default.
- W3215093302 citedByCount "2" @default.
- W3215093302 countsByYear W32150933022012 @default.
- W3215093302 crossrefType "journal-article" @default.
- W3215093302 hasAuthorship W3215093302A5020085889 @default.
- W3215093302 hasAuthorship W3215093302A5033093547 @default.
- W3215093302 hasAuthorship W3215093302A5090309356 @default.
- W3215093302 hasBestOaLocation W32150933021 @default.
- W3215093302 hasConcept C119857082 @default.
- W3215093302 hasConcept C154945302 @default.
- W3215093302 hasConcept C164155591 @default.
- W3215093302 hasConcept C2781045450 @default.
- W3215093302 hasConcept C38652104 @default.
- W3215093302 hasConcept C41008148 @default.
- W3215093302 hasConcept C41608201 @default.
- W3215093302 hasConcept C80444323 @default.
- W3215093302 hasConcept C84525736 @default.
- W3215093302 hasConcept C95623464 @default.
- W3215093302 hasConceptScore W3215093302C119857082 @default.
- W3215093302 hasConceptScore W3215093302C154945302 @default.
- W3215093302 hasConceptScore W3215093302C164155591 @default.
- W3215093302 hasConceptScore W3215093302C2781045450 @default.
- W3215093302 hasConceptScore W3215093302C38652104 @default.
- W3215093302 hasConceptScore W3215093302C41008148 @default.
- W3215093302 hasConceptScore W3215093302C41608201 @default.
- W3215093302 hasConceptScore W3215093302C80444323 @default.
- W3215093302 hasConceptScore W3215093302C84525736 @default.
- W3215093302 hasConceptScore W3215093302C95623464 @default.
- W3215093302 hasFunder F4320335254 @default.
- W3215093302 hasIssue "5" @default.
- W3215093302 hasLocation W32150933021 @default.
- W3215093302 hasLocation W32150933022 @default.
- W3215093302 hasLocation W32150933023 @default.
- W3215093302 hasOpenAccess W3215093302 @default.
- W3215093302 hasPrimaryLocation W32150933021 @default.
- W3215093302 hasRelatedWork W1470425429 @default.
- W3215093302 hasRelatedWork W2940336242 @default.
- W3215093302 hasRelatedWork W3127425528 @default.
- W3215093302 hasRelatedWork W4205478082 @default.
- W3215093302 hasRelatedWork W4281385048 @default.
- W3215093302 hasRelatedWork W4308191010 @default.
- W3215093302 hasRelatedWork W4313001487 @default.
- W3215093302 hasRelatedWork W4318350883 @default.
- W3215093302 hasRelatedWork W4328134586 @default.
- W3215093302 hasRelatedWork W4361795583 @default.
- W3215093302 hasVolume "111" @default.
- W3215093302 isParatext "false" @default.
- W3215093302 isRetracted "false" @default.
- W3215093302 magId "3215093302" @default.
- W3215093302 workType "article" @default.