FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3198519141> ?p ?o ?g. }

• W3198519141 endingPage "545" @default.
• W3198519141 startingPage "519" @default.
• W3198519141 abstract "Abstract Verification has emerged as a means to provide formal guarantees on learning-based systems incorporating neural network before using them in safety-critical applications. This paper proposes a new verification approach for deep neural networks (DNNs) with piecewise linear activation functions using reachability analysis. The core of our approach is a collection of reachability algorithms using star sets (or shortly, stars), an effective symbolic representation of high-dimensional polytopes. The star-based reachability algorithms compute the output reachable sets of a network with a given input set before using them for verification. For a neural network with piecewise linear activation functions, our approach can construct both exact and over-approximate reachable sets of the neural network. To enhance the scalability of our approach, a star set is equipped with an outer-zonotope (a zonotope over-approximation of the star set) to quickly estimate the lower and upper bounds of an input set at a specific neuron to determine if splitting occurs at that neuron. This zonotope pre-filtering step reduces significantly the number of linear programming optimization problems that must be solved in the analysis, and leads to a reduction in computation time, which enhances the scalability of the star set approach. Our reachability algorithms are implemented in a software prototype called the neural network verification tool, and can be applied to problems analyzing the robustness of machine learning methods, such as safety and robustness verification of DNNs. Our experiments show that our approach can achieve runtimes twenty to 1400 times faster than Reluplex, a satisfiability modulo theory-based approach. Our star set approach is also less conservative than other recent zonotope and abstract domain approaches." @default.
• W3198519141 created "2021-09-13" @default.
• W3198519141 creator A5005299838 @default.
• W3198519141 creator A5005381976 @default.
• W3198519141 creator A5021939745 @default.
• W3198519141 creator A5023289374 @default.
• W3198519141 creator A5033330108 @default.
• W3198519141 creator A5051151589 @default.
• W3198519141 creator A5056925300 @default.
• W3198519141 creator A5067901159 @default.
• W3198519141 creator A5090573416 @default.
• W3198519141 date "2021-08-01" @default.
• W3198519141 modified "2023-09-26" @default.
• W3198519141 title "Verification of piecewise deep neural networks: a star set approach with zonotope pre-filter" @default.
• W3198519141 cites W1511213320 @default.
• W3198519141 cites W2160815625 @default.
• W3198519141 cites W2165073069 @default.
• W3198519141 cites W2243397390 @default.
• W3198519141 cites W2543296129 @default.
• W3198519141 cites W2565516711 @default.
• W3198519141 cites W2592929672 @default.
• W3198519141 cites W2594877703 @default.
• W3198519141 cites W2734705105 @default.
• W3198519141 cites W2791251367 @default.
• W3198519141 cites W2794609696 @default.
• W3198519141 cites W2896500468 @default.
• W3198519141 cites W2900153411 @default.
• W3198519141 cites W2905299363 @default.
• W3198519141 cites W2955364893 @default.
• W3198519141 cites W2957311447 @default.
• W3198519141 cites W2963054787 @default.
• W3198519141 cites W2974147052 @default.
• W3198519141 cites W3042471904 @default.
• W3198519141 cites W3043655933 @default.
• W3198519141 cites W3046662910 @default.
• W3198519141 cites W3186627997 @default.
• W3198519141 doi "https://doi.org/10.1007/s00165-021-00553-4" @default.
• W3198519141 hasPublicationYear "2021" @default.
• W3198519141 type Work @default.
• W3198519141 sameAs 3198519141 @default.
• W3198519141 citedByCount "6" @default.
• W3198519141 countsByYear W31985191412022 @default.
• W3198519141 countsByYear W31985191412023 @default.
• W3198519141 crossrefType "journal-article" @default.
• W3198519141 hasAuthorship W3198519141A5005299838 @default.
• W3198519141 hasAuthorship W3198519141A5005381976 @default.
• W3198519141 hasAuthorship W3198519141A5021939745 @default.
• W3198519141 hasAuthorship W3198519141A5023289374 @default.
• W3198519141 hasAuthorship W3198519141A5033330108 @default.
• W3198519141 hasAuthorship W3198519141A5051151589 @default.
• W3198519141 hasAuthorship W3198519141A5056925300 @default.
• W3198519141 hasAuthorship W3198519141A5067901159 @default.
• W3198519141 hasAuthorship W3198519141A5090573416 @default.
• W3198519141 hasBestOaLocation W31985191411 @default.
• W3198519141 hasConcept C104317684 @default.
• W3198519141 hasConcept C11413529 @default.
• W3198519141 hasConcept C136643341 @default.
• W3198519141 hasConcept C154945302 @default.
• W3198519141 hasConcept C185592680 @default.
• W3198519141 hasConcept C41008148 @default.
• W3198519141 hasConcept C48044578 @default.
• W3198519141 hasConcept C50644808 @default.
• W3198519141 hasConcept C55493867 @default.
• W3198519141 hasConcept C63479239 @default.
• W3198519141 hasConcept C77088390 @default.
• W3198519141 hasConcept C80444323 @default.
• W3198519141 hasConceptScore W3198519141C104317684 @default.
• W3198519141 hasConceptScore W3198519141C11413529 @default.
• W3198519141 hasConceptScore W3198519141C136643341 @default.
• W3198519141 hasConceptScore W3198519141C154945302 @default.
• W3198519141 hasConceptScore W3198519141C185592680 @default.
• W3198519141 hasConceptScore W3198519141C41008148 @default.
• W3198519141 hasConceptScore W3198519141C48044578 @default.
• W3198519141 hasConceptScore W3198519141C50644808 @default.
• W3198519141 hasConceptScore W3198519141C55493867 @default.
• W3198519141 hasConceptScore W3198519141C63479239 @default.
• W3198519141 hasConceptScore W3198519141C77088390 @default.
• W3198519141 hasConceptScore W3198519141C80444323 @default.
• W3198519141 hasIssue "4-5" @default.
• W3198519141 hasLocation W31985191411 @default.
• W3198519141 hasOpenAccess W3198519141 @default.
• W3198519141 hasPrimaryLocation W31985191411 @default.
• W3198519141 hasRelatedWork W1486100501 @default.
• W3198519141 hasRelatedWork W1525643724 @default.
• W3198519141 hasRelatedWork W1534259281 @default.
• W3198519141 hasRelatedWork W2067938758 @default.
• W3198519141 hasRelatedWork W2302028273 @default.
• W3198519141 hasRelatedWork W2364921833 @default.
• W3198519141 hasRelatedWork W2368796975 @default.
• W3198519141 hasRelatedWork W2573691487 @default.
• W3198519141 hasRelatedWork W381935118 @default.
• W3198519141 hasRelatedWork W4299419634 @default.
• W3198519141 hasVolume "33" @default.
• W3198519141 isParatext "false" @default.
• W3198519141 isRetracted "false" @default.
• W3198519141 magId "3198519141" @default.
• W3198519141 workType "article" @default.