New Threat on Formal Verification for Neural Networks: Example and Fault Tolerance

Augustin Viot (1, 2, 3) , Benjamin Lussier (2) , Walter Schön (2) , Armando Tacchella (1) , Stéphane Geronimi (3)

1 DIBRIS - Dipartimento di Informatica, Bioingegneria, Robotica e Ingegneria dei Sistemi [Genova]
2 Heudiasyc - Heuristique et Diagnostic des Systèmes Complexes [Compiègne]
3 Stellantis France

Domaines: Machine Learning [stat.ML]; Intelligence artificielle [cs.AI]; Robotique [cs.RO]; Performance et fiabilité [cs.PF]; Réseau de neurones [cs.NE]

Augustin Viot, Benjamin Lussier, Walter Schön, Armando Tacchella, Stéphane Geronimi. New Threat on Formal Verification for Neural Networks: Example and Fault Tolerance. 11th IFAC Symposium on Fault Detection, Supervision and Safety for Technical Processes (SAFEPROCESS 2022), Jun 2022, Pafos, Cyprus. pp.623-630, ⟨10.1016/j.ifacol.2022.07.197⟩. ⟨hal-03823896⟩