Abstraction of Neural Network
Verification of NN is crucial with their rise in safety-critical applications, as neural networks can be fooled by applying small perturbations to the input. However, we face heavy scalability issues due to the size of modern architectures. To this end, we provide abstraction frameworks to reduce the size of the NN while keeping guarantees. (DeepAbstract, LiNNA)
Team
Tools
Publications
|
2023 |
Syntactic vs Semantic Linear Abstraction and Refinement of Neural Networks
ATVA 2023 |
|
2020 |
DeepAbstract: Neural Network Abstraction for Accelerating Verification
ATVA 2020 |