survey-neural-network-verification

List of papers about neural network verificcation.

Review

• A Review of Formal Methods applied to Machine Learning.

Adversarial Examples

• Explaining and Harnessing Adversarial Examples.

Complete verification

• Evaluating Robustness of Neural Networks with Mixed Integer Programming.
• Branch and Bound for Piecewise Linear Neural Network Verification.
• An Unified View of Piecewise Linear Neural Network Verification.
• Reluplex: An Efficient SMT Solver for Verifying Deep Neural Networks.

Incomplete verification

Abstract interpretation-based approach

• AI2: Safety and Robustness Certification of Neural Networks with Abstract Interpretation.
• Fast and Effective Robustness Certification.
• An abstract domain for certifying neural networks.
• Boosting Robustness Certification of Neural Networks.
• Beyond the Single Neuron Convex Barrier for Nerual Network Certification.
• PRIMA: General and Precise Neural Network Certification via Scalable Convex Hull Approximates.

Decomposition-based approach

• A Dual Approach to Scalable Verification of Deep Networks.
• Lagrangian Decomposition for Neural Network Verification.
• Improved Branch and Bound for Neural Network Verification via Lagrangian Decomposition.

Different Perturbations

• Certifying Geometric Robustness of Neural Networks.

Reuse proof effort

• Shared Certificates for Neural Network Verification.

VNNCOMP reports