signSGD via Zeroth-Order Oracle

Authors: Sijia Liu, Pin-Yu Chen, Xiangyi Chen, Mingyi Hong

ICLR 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our empirical evaluations on image classification datasets MNIST and CIFAR-10 demonstrate the superior performance of ZO-sign SGD on the generation of adversarial examples from black-box neural networks.
Researcher Affiliation Collaboration Sijia Liu Pin-Yu Chen Xiangyi Chen Mingyi Hong MIT-IBM Watson AI Lab, IBM Research University of Minnesota, Twin Cities
Pseudocode Yes Algorithm 1 Generic sign-based gradient descent
Open Source Code No The paper does not include an unambiguous statement of releasing its own source code or a direct link to a repository for the work described.
Open Datasets Yes Our empirical evaluations on image classification datasets MNIST and CIFAR-10
Dataset Splits No The paper mentions training samples and testing samples (n = 2000 training, 200 testing) for the synthetic dataset, but does not explicitly specify a validation set or general data split percentages for reproduction across all datasets.
Hardware Specification No The paper does not specify any particular hardware (e.g., GPU models, CPU types, or cloud computing instances with specifications) used for running the experiments.
Software Dependencies No The paper does not explicitly list software dependencies with specific version numbers (e.g., 'Python 3.8, PyTorch 1.9, and CUDA 11.1').
Experiment Setup Yes We find the best constant learning rate for algorithms via a greedy search over η [0.001, 0.1] (see Appendix 8.1 for more details), and we choose the smoothing parameter µ = 10/ Td. Unless specified otherwise, let b = q = 10, T = 5000 and d = 100. In our experiment, we set c = 1 for MNIST and c = 0.1 for CIFAR-10. We also set the same parameters for each method, i.e., µ = 0.01, q = 9, and δ = 0.05 for MNIST and δ = 0.0005 for CIFAR-10, to accommodate to the dimension factor d.