PyRAT: A Tool for Verifying the Safety and Robustness of Neural Networks Using Abstract Interpretation
PyRAT is a Python-based tool that leverages abstract interpretation techniques to verify the safety and robustness of neural networks, offering various abstract domains and optimization strategies to provide formal guarantees about network behavior.
Differential Verification for Neural Network Equivalence: A Zonotope-based Approach with Confidence Guarantees
This paper proposes a novel approach to verifying the equivalence of neural networks, particularly after pruning and retraining, using differential verification with Zonotopes and introducing a new confidence-based equivalence property for enhanced guarantees.
The Undecidability of the Reachability Problem for Neural-Network Control Systems
Determining whether a neural-network control system can reach a specific state from a set of initial states is undecidable, even for simple systems.
DEEPCDCL: A Novel Neural Network Verification Framework Based on CDCL Algorithm
DEEPCDCL introduces an efficient neural network verification framework based on the Conflict-Driven Clause Learning (CDCL) algorithm, aiming to enhance speed and accuracy in verification processes.
© 2024 by Linnk AI