Mirror Descent Maximizes Generalized Margin and Can Be Implemented Efficiently
Authors: Haoyuan Sun, Kwangjun Ahn, Christos Thrampoulidis, Navid Azizan
NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Using comprehensive experiments with both linear and deep neural network models, we demonstrate that p-GD can noticeably affect the structure and the generalization performance of the learned models. |
| Researcher Affiliation | Academia | Haoyuan Sun MIT haoyuans@mit.edu Kwangjun Ahn MIT kjahn@mit.edu Christos Thrampoulidis UBC cthrampo@ece.ubc.ca Navid Azizan MIT azizan@mit.edu |
| Pseudocode | No | The paper provides mathematical update rules and mentions 'code snippets in Appendix D', but it does not include a formally labeled 'Pseudocode' or 'Algorithm' block. |
| Open Source Code | Yes | Did you include the code, data, and instructions needed to reproduce the main experimental results (either in the supplemental material or as a URL)? [Yes] They are included in the supplemental material. |
| Open Datasets | Yes | Specifically, we perform a set of experiments on the CIFAR-10 dataset [Krizhevsky et al., 2009]. |
| Dataset Splits | No | The paper states 'We gave an overview of our training setup in Section 4 and the full details are given in Appendix E.', but the provided text does not explicitly detail the training/validation/test dataset splits with percentages or sample counts. |
| Hardware Specification | No | The acknowledgement section mentions 'MIT Super Cloud and Lincoln Laboratory Supercomputing Center for providing computing resources', but these are general names for computing centers and do not specify particular hardware details like GPU or CPU models. The paper checklist states details are in Appendix E, which is not provided. |
| Software Dependencies | No | The paper does not provide specific version numbers for any software dependencies used in the experiments. |
| Experiment Setup | Yes | We used fixed step size of 10-4 and ran 250 thousand iterations for different p s. |