Training for Verification: Increasing Neuron Stability to Scale DNN Verification.

AbstractWith the growing use of deep neural networks(DNN) in mission and safety-critical applications, there is an increasing interest in DNN verification. Unfortunately, increasingly complex network structures, non-linear behavior, and high-dimensional input spaces combine to make DNN verification computationally challenging. Despite tremendous advances, DNN verifiers are still challenged to scale to large verification problems. In this work, we explore how the number of stable neurons under the precondition of a specification gives rise to verification complexity. We examine prior work on the problem, adapt it, and develop several novel approaches to increase stability. We demonstrate that neuron stability can be increased substantially without compromising model accuracy and this yields a multi-fold improvement in DNN verifier performance.