Set-based training of neural barrier certificates uses a loss function that encodes all safety properties so that zero loss formally proves the certificate is valid, collapsing iterative training and verification into one procedure.
Multilayer feedforward networks with a nonpolynomial activation function can approximate any function
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.SY 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Set-Based Training of Neural Barrier Certificates for Safety Verification of Dynamical Systems
Set-based training of neural barrier certificates uses a loss function that encodes all safety properties so that zero loss formally proves the certificate is valid, collapsing iterative training and verification into one procedure.