ISO-24029

Artificial intelligence — Assessment of the robustness of neural networks

Multi-part standard for assessing the robustness of neural networks. Covers formal methods, statistical approaches, and perturbation-based testing for evaluating how neural networks respond to adversarial inputs and distributional shifts.

Document
ISO/IEC 24029 (series, 2021)
URL
https://www.iso.org/standard/77609.html
Label
Standard
Keywords
neural networks robustness adversarial inputs formal methods perturbation testing

Domain: Artificial Intelligence and Machine Learning · Standard