PV3 Trainifychecks properties of a deep reinforcement learning systems

Application domain/field

Type of tool

Framework for training & verification of deep reinforcement learning systems

Expected input


? (seems to all be hardcoded in Python)

Expected output

Whether the property holds or not.


The tool builds a finite abstract state space using the trained neural network.
DNN Model checking


- Repository: https://github.com/aptx4869tjx/RL_verification

Related papers

TRAINIFY: A CEGAR-Driven Training and Verification Framework for Safe Deep Reinforcement Learning (CAV 2022)

Last publication date

7 August 2022

ProVerB specific

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