Verification

Fairify: Fairness Verification of Neural Networks
We proposed Fairify, an approach to make individual fairness verification tractable for the developers. The key idea is that many neurons in the NN always remain inactive when a smaller part of the input domain is considered. So, Fairify leverages white-box access to the models in production and then apply formal analysis based pruning.
Sumon Biswas, Hridesh Rajan
Cite Code DOI PDF
Fairify: Fairness Verification of Neural Networks
Safety Assurance of Predictive Systems
We built abstractions of ML systems and inferred preconditions to provide assurance in safety-critical predictions.
Safety Assurance of Predictive Systems
Verifying Neural Networks for Individual Fairness
A modular approach to formally verify neural networks. We specified individual property for SMT solver and verified fairness for specific subpopulations.
Verifying Neural Networks for Individual Fairness