2026-07-22T00:09:45Z https://drops.dagstuhl.de/oai

oai:drops-oai.dagstuhl.de:10903 2024-03-06T10:47:09Z ddc:004 open_access

Safety Verification for Deep Neural Networks with Provable Guarantees (Invited Paper) Kwiatkowska, Marta Z. Neural networks robustness formal verification Bayesian neural networks Computing systems are becoming ever more complex, increasingly often incorporating deep learning components. Since deep learning is unstable with respect to adversarial perturbations, there is a need for rigorous software development methodologies that encompass machine learning. This paper describes progress with developing automated verification techniques for deep neural networks to ensure safety and robustness of their decisions with respect to input perturbations. This includes novel algorithms based on feature-guided search, games, global optimisation and Bayesian methods. Schloss Dagstuhl – Leibniz-Zentrum für Informatik Marta Z. Kwiatkowska 2019 Is Part Of LIPIcs, Volume 140, 30th International Conference on Concurrency Theory (CONCUR 2019) InProceedings Text doc-type:ResearchArticle publishedVersion application/pdf doi:10.4230/LIPIcs.CONCUR.2019.1 urn:nbn:de:0030-drops-109036 https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.CONCUR.2019.1 eng https://creativecommons.org/licenses/by/3.0/legalcode