Tight Certification of Adversarially Trained Neural Networks via Nonconvex Low-Rank Semidefinite Relaxations.
Hong-Ming ChiuRichard Y. ZhangPublished in: ICML (2023)
Keyphrases
- low rank
- semidefinite
- convex optimization
- convex relaxation
- interior point methods
- convex sets
- semidefinite programming
- kernel matrix
- lower bound
- primal dual
- total variation
- matrix factorization
- missing data
- neural network
- high order
- higher dimensional
- singular value decomposition
- collaborative filtering
- genetic algorithm