Safety Verification and Robustness Analysis of Neural Networks via Quadratic Constraints and Semidefinite Programming

Certifying the safety or robustness of neural networks against input uncertainties and adversarial attacks is an emerging challenge in the area of safe machine learning and control. To provide such a guarantee, one must be able to bound the output of neural networks when their input changes within a...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:IEEE transactions on automatic control Ročník 67; číslo 1; s. 1 - 15
Hlavní autori: Fazlyab, Mahyar, Morari, Manfred, Pappas, George J.
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: New York IEEE 01.01.2022
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Predmet:
Biological neural networks
Convex optimization
deep neural networks
Machine learning
Neural networks
Perturbation methods
Programming
Robustness
Robustness (mathematics)
robustness analysis
Safety
safety verification
Semidefinite programming
semidefinite programming (SDP)
Uncertainty
Verification
ISSN:0018-9286, 1558-2523
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:Certifying the safety or robustness of neural networks against input uncertainties and adversarial attacks is an emerging challenge in the area of safe machine learning and control. To provide such a guarantee, one must be able to bound the output of neural networks when their input changes within a bounded set. In this article, we propose a semidefinite programming (SDP) framework to address this problem for feed-forward neural networks with general activation functions and input uncertainty sets. Our main idea is to abstract various properties of activation functions (e.g., monotonicity, bounded slope, bounded values, and repetition across layers) with the formalism of quadratic constraints. We then analyze the safety properties of the abstracted network via the S -procedure and SDP. Our framework spans the tradeoff between conservatism and computational efficiency and applies to problems beyond safety verification. We evaluate the performance of our approach via numerical problem instances of various sizes.
Bibliografia:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0018-9286
1558-2523
DOI:10.1109/TAC.2020.3046193