Deep Learning Toolbox Verification Library

Verify and test robustness of deep learning networks, deploy with confidence

Deep Learning Toolbox Verification Library allows you to verify and test properties of deep learning networks, and deploy these models with confidence.

Use this library to:

• Verify network robustness to adversarial examples (Since R2022b)
• Estimate how sensitive the network predictions are to input perturbation (Since R2022b)
• Verify network properties in parellel with multiple GPU and CPU support (Since R2024a; Library Version 24.1.1)
• Verify branched networks (Since R2024b; Library Version 24.2.2)
• Explain object detection network predictions using D-RISE (Since R2024a)
• Create a distribution discriminator that separates data into in- and out-of-distribution (Since R2023a)
• Runtime Monitoring: detect out-of-distribution (ODD) data in neural networks (Since R2023a)
• Runtime Monitoring: generate C/C++ and CUDA code for out-of-distribution detection (Since R2023a)

Please refer to the documentation here: https://www.mathworks.com/help/deeplearning/verification.html

If you have download or installation problems, please contact Technical Support: https://www.mathworks.com/support/contact_us.html