A Semidefinite Relaxation Based Branch-and-Bound Method for Tight Neural Network Verification.
Jianglin LanBenedikt BrücknerAlessio LomuscioPublished in: AAAI (2023)
Keyphrases
- semidefinite
- branch and bound method
- lower bound
- neural network
- branch and bound
- upper bound
- branch and bound algorithm
- semidefinite programming
- mixed integer programming
- convex relaxation
- reduce the search space
- optimal configuration
- higher dimensional
- sufficient conditions
- feasible solution
- np hard
- interior point methods
- lagrangian relaxation
- worst case
- optimal solution
- convex sets
- objective function
- search space
- lot sizing
- column generation
- search algorithm
- combinatorial optimization
- finite dimensional
- computational complexity
- bayesian networks