English

Property-Directed Verification of Recurrent Neural Networks

Machine Learning 2020-09-23 v1 Artificial Intelligence Formal Languages and Automata Theory Machine Learning

Abstract

This paper presents a property-directed approach to verifying recurrent neural networks (RNNs). To this end, we learn a deterministic finite automaton as a surrogate model from a given RNN using active automata learning. This model may then be analyzed using model checking as verification technique. The term property-directed reflects the idea that our procedure is guided and controlled by the given property rather than performing the two steps separately. We show that this not only allows us to discover small counterexamples fast, but also to generalize them by pumping towards faulty flows hinting at the underlying error in the RNN.

Keywords

Cite

@article{arxiv.2009.10610,
  title  = {Property-Directed Verification of Recurrent Neural Networks},
  author = {Igor Khmelnitsky and Daniel Neider and Rajarshi Roy and Benoît Barbot and Benedikt Bollig and Alain Finkel and Serge Haddad and Martin Leucker and Lina Ye},
  journal= {arXiv preprint arXiv:2009.10610},
  year   = {2020}
}