Skip to content

Latest commit

 

History

History
 
 

neural_mip_solving

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 

Neural MIP Solving - NN Verification Dataset

This is the “Neural Network Verification” dataset used in the paper

Solving Mixed Integer Programs Using Neural Networks (Nair et al., 2020).

It contains a set of mixed integer programs (MIPs) for the problem of verifying a neural network’s robustness to perturbations to its inputs. The MIP formulation is described in the paper On the Effectiveness of Interval Bound Propagation for Training Verifiably Robust Models (Gowal et al., 2018).

This dataset corresponds to MIPs defined for verifying a neural network with the architecture labelled as “small” in Table 1 of Gowal et al., 2018, and trained on the MNIST image dataset. The code used to train the neural network to be verified is available at https://github.com/deepmind/interval-bound-propagation. The MIPs are split into the same training, validation, and test sets as that in Nair et al., 2020.

Dataset Location

The dataset is available in the following link

Dataset Metadata

The following table is necessary for this dataset to be indexed by search engines such as Google Dataset Search.

property value
name Neural Network Verification Dataset
url
sameAs https://github.com/deepmind/deepmind-research/tree/master/neural_mip_solving
description This dataset contains a set of mixed integer programs (MIPs) for the problem of verifying a neural network’s robustness to perturbations of its inputs. The MIPs are encoded in LP format.
license https://creativecommons.org/licenses/by/4.0/legalcode
provider
property value
name DeepMind
sameAs https://en.wikipedia.org/wiki/DeepMind
citation https://arxiv.org/abs/2012.13349

Citing this work

If you use this dataset in your work, we ask you to cite this paper:

@misc{nair2020solving,
      title={Solving Mixed Integer Programs Using Neural Networks},
      author={Vinod Nair and Sergey Bartunov and Felix Gimeno and Ingrid von Glehn and Pawel Lichocki and Ivan Lobov and Brendan O'Donoghue and Nicolas Sonnerat and Christian Tjandraatmadja and Pengming Wang and Ravichandra Addanki and Tharindi Hapuarachchi and Thomas Keck and James Keeling and Pushmeet Kohli and Ira Ktena and Yujia Li and Oriol Vinyals and Yori Zwols},
      year={2020},
      eprint={2012.13349},
      archivePrefix={arXiv},
      primaryClass={math.OC}
}

License

This dataset is made available under the terms of the Creative Commons Attribution 4.0 International (CC BY 4.0) license.

You can find details at: https://creativecommons.org/licenses/by/4.0/legalcode

Disclaimer

This is not an officially supported Google product.