PV1 ⊧ PROVER: Polyhedral Robustness Verifier of RNNsruns 100 tests through a neural network, testing its noise robustness

Verifier for recurrent neural networks (RNNs).

Application domain/field

Internals

The goal of PROVER is to establish the robustness of recurrent neural networks against noise-induced perturbations. They specifically focus on Long Short-Term Memory (LSTMs). Uses Gurobi
Neural network

Links

Repository: https://github.com/eth-sri/prover

Related papers

Scalable Polyhedral Verification of Recurrent Neural Networks (CAV '21)

Last publication date

15 July 2021

Related tools

POPQORN

ProVerB specific


ProVerB is a part of SLEBoK. Last updated: February 2023.