Verifying Learning Artificial Intelligence Systems

Projet: Recherche

Détails du projet

Description

With great power comes great responsibility" is a cliché that perfectly applies to artificial
intelligence (AI). As technology is progressing faster than ever before towards software with
more and more abilities, adaptation and autonomy, the risks are becoming increasingly
apparent. As a matter of fact, humans are still lacking the tools to systematically obtain strong
guarantees - about their safety, privacy, etc. - from the intelligent software that serves them.
Quite paradoxically, but not so surprisingly, better human control over AI could come from
intelligent software, and more specifically, from automated verification software. Verification
is a scientific discipline within computer science that devises automated methods to improve
the dependability of computer-based systems. It does so by exploiting mathematicallygrounded
techniques such as model checking, theorem proving and model-based testing.
Although automated verification has made remarkable progress during the past decades, its
application is still limited to traditional software that does not use AI or machine learning
technology. This project aims to remove these limitations by investigating the logical and
probabilistic underpinnings of both verification, machine learning and AI. The expected results
are new theories, formalisms and algorithms that will constitute the foundations for a new
generation of AI-ready verification methods.
AcronymeVeriLearn
statutFini
Les dates de début/date réelle1/01/1831/12/22

Attachement à un institut de recherche reconnus à l'UNAMUR

  • NADI

Empreinte digitale

Explorez les thèmes de recherche abordés par ce projet. Ces libellés sont générés sur la base des prix/subventions sous-jacents. Ensemble, ils forment une empreinte digitale unique.
  • Towards Feature-based ML-enabled Behaviour Location

    Fortz, S., Temple, P., Devroey, X. & Perrouin, G., 7 févr. 2024, Proceedings of the 18th International Working Conference on Variability Modelling of Software-Intensive Systems (VaMoS 2024). Bern, Switzerland: ACM Press, 3 p. (Proceedings of the 18th International Working Conference on Variability Modelling of Software-Intensive Systems).

    Résultats de recherche: Contribution dans un livre/un catalogue/un rapport/dans les actes d'une conférenceArticle dans les actes d'une conférence/un colloque

    Accès ouvert
    File
    2 Téléchargements (Pure)
  • Global Explanations with Decision Rules: a Co-learning Approach

    Nanfack, G., Temple, P. & Frénay, B., 27 juil. 2021, Proceedings of the Thirty-Seventh Conference on Uncertainty in Artificial Intelligence. de Campos, C. & Maathuis, M. H. (eds.). MLResearch Press, p. 589-599 11 p. (Proceedings of Machine Learning Research ; Vol 161).

    Résultats de recherche: Contribution dans un livre/un catalogue/un rapport/dans les actes d'une conférenceArticle dans les actes d'une conférence/un colloque

    Accès ouvert
    File
    79 Téléchargements (Pure)
  • VaryMinions: Leveraging RNNs to Identify Variants in Event Logs

    Fortz, S., Temple, P., DEVROEY, X., HEYMANS, P. & PERROUIN, GILLES., 2021, 5th International Workshop on Machine Learning Techniques for Software Quality Evolution. ACM Press

    Résultats de recherche: Contribution dans un livre/un catalogue/un rapport/dans les actes d'une conférenceArticle dans les actes d'une conférence/un colloque

    Accès ouvert
    File
    122 Téléchargements (Pure)