Probabilistic Verification of Fairness Properties via Concentration
This addresses the critical need to prevent discrimination in automated decision-making systems, though it is incremental as it builds on existing verification techniques with probabilistic guarantees.
The authors tackled the problem of verifying fairness in machine learning systems used for legal and financial decisions by designing a scalable algorithm based on adaptive concentration inequalities, which they implemented in VeriFair to handle models like a deep recurrent neural network over 100,000 times larger than previously verified ones with extremely small error probabilities.
As machine learning systems are increasingly used to make real world legal and financial decisions, it is of paramount importance that we develop algorithms to verify that these systems do not discriminate against minorities. We design a scalable algorithm for verifying fairness specifications. Our algorithm obtains strong correctness guarantees based on adaptive concentration inequalities; such inequalities enable our algorithm to adaptively take samples until it has enough data to make a decision. We implement our algorithm in a tool called VeriFair, and show that it scales to large machine learning models, including a deep recurrent neural network that is more than five orders of magnitude larger than the largest previously-verified neural network. While our technique only gives probabilistic guarantees due to the use of random samples, we show that we can choose the probability of error to be extremely small.