Globally-Robust Neural Networks

Authors: Klas Leino, Zifan Wang, Matt Fredrikson

ICML 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we present an empirical evaluation of our method.
Researcher Affiliation Academia Klas Leino 1 Zifan Wang 1 Matt Fredrikson 1 1Carnegie Mellon University, Pittsburgh, Pennsylvania, USA.
Pseudocode No No pseudocode or algorithm blocks were found in the paper. The methods are described in prose and mathematical formulations.
Open Source Code Yes An implementation of our approach is available on Git Hub1. 1Code available at https://github.com/klasleino/gloro
Open Datasets Yes MNIST (Le Cun et al., 2010), CIFAR-10 (Krizhevsky, 2009) and Tiny-Imagenet (Le & Yang, 2015)
Dataset Splits No The paper mentions training on datasets like MNIST, CIFAR-10, and Tiny-Imagenet and evaluating on the "entire test set," but does not explicitly provide the specific percentages or methodology used for training, validation, and test data splits.
Hardware Specification Yes All timings were taken on a machine using a Geforce RTX 3080 accelerator, 64 GB memory, and Intel i9 10850K CPU, with the exception of those for the KW (Wong et al., 2018) method, which were taken on a Titan RTX card for toolkit compatibility reasons.
Software Dependencies No The paper mentions using ART (Nicolae et al., 2019) for PGD attacks but does not provide specific version numbers for software dependencies like Python, deep learning frameworks (e.g., PyTorch, TensorFlow), or CUDA.
Experiment Setup Yes Further details on the precise hyperparameters used for training and attacks, the process for obtaining these parameters, and the network architectures are provided in Appendix B.