An Abstraction-Based Framework for Neural Network Verification

Simple item page
dc.contributor.authorGottschlich, Justin E
dc.contributor.authorGottschlich, Justin E
dc.contributor.authorKatz, Guy
dc.date2023-05-18T00:14:22.000
dc.date.accessioned2023-05-22T13:06:47Z
dc.date.available2023-05-22T13:06:47Z
dc.date.issued2020-01-01
dc.date.submitted2020-12-18T13:16:19-08:00
dc.description.abstractDeep neural networks are increasingly being used as controllers for safety-critical systems. Because neural networks are opaque, certifying their correctness is a significant challenge. To address this issue, several neural network verification approaches have recently been proposed. However, these approaches afford limited scalability, and applying them to large networks can be challenging. In this paper, we propose a framework that can enhance neural network verification techniques by using over-approximation to reduce the size of the network—thus making it more amenable to verification. We perform the approximation such that if the property holds for the smaller (abstract) network, it holds for the original as well. The over-approximation may be too coarse, in which case the underlying verification tool might return a spurious counterexample. Under such conditions, we perform counterexample-guided refinement to adjust the approximation, and then repeat the process. Our approach is orthogonal to, and can be integrated with, many existing verification techniques. For evaluation purposes, we integrate it with the recently proposed Marabou framework, and observe a significant improvement in Marabou’s performance. Our experiments demonstrate the great potential of our approach for verifying larger neural networks.
dc.identifier.urihttps://repository.upenn.edu/handle/20.500.14332/8491
dc.legacy.articleid1004
dc.legacy.fulltexturlhttps://repository.upenn.edu/cgi/viewcontent.cgi?article=1004&context=cps_machine_programming&unstamped=1
dc.source.issue5
dc.source.journalMachine Programming
dc.source.statuspublished
dc.titleAn Abstraction-Based Framework for Neural Network Verification
dc.typeBook
digcom.contributor.authorElboher, Yizhak Y
digcom.contributor.authorisAuthorOfPublication|email:gojustin@cis.upenn.edu|institution:Intel|Gottschlich, Justin E
digcom.contributor.authorKatz, Guy
digcom.identifiercps_machine_programming/5
digcom.identifier.contextkey20688866
digcom.identifier.submissionpathcps_machine_programming/5
digcom.typebook
dspace.entity.typePublication
relation.isAuthorOfPublication5cbcf403-a558-4c1c-aa8a-d700e3d50679
relation.isAuthorOfPublication.latestForDiscovery5cbcf403-a558-4c1c-aa8a-d700e3d50679
upenn.schoolDepartmentCenterMachine Programming
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
Elboher2020_Chapter_AnAbstraction_BasedFrameworkFo.pdf
Size:
1.32 MB
Format:
Adobe Portable Document Format
Download
Collection
Books