Convex Calibrated Surrogates for the Multi-Label F-Measure
Authors: Mingyuan Zhang, Harish Guruprasad Ramaswamy, Shivani Agarwal
ICML 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our experiments confirm our theoretical findings. (Abstract) and We conducted two sets of experiments to evaluate our algo rithm. (Section 7. Experiments) |
| Researcher Affiliation | Academia | 1Department of Computer and Information Science, University of Pennsylvania, Philadelphia, PA, USA 2Department of Com puter Science and Engineering, Indian Institute of Technology Madras, Chennai, India. |
| Pseudocode | Yes | Algorithm 1 Surrogate risk minimization algorithm for multi-label Fβ -measure (Section 4) |
| Open Source Code | No | The paper does not provide an explicit statement or link for the open-source code of the methodology described. |
| Open Datasets | Yes | We evaluated the perfor mance of our algorithm on various benchmark multi-label data sets drawn from the Mulan repository6. Details of the data sets are provided in Table 1. Footnote 6: 6http://mulan.sourceforge.net/datasets-mlc.html (Section 7.2) |
| Dataset Splits | Yes | All the data sets come with prescribed train/test splits. and Regularization parameters (for regularized logis tic regression in our algorithm, EFP, and BR; and for the margin-based objective in LIMO) were chosen by 5-fold cross-validation on the training set from {10 4 , . . . , 103} (Section 7.2) |
| Hardware Specification | No | The paper does not provide specific hardware details used for running its experiments. |
| Software Dependencies | No | The paper does not provide specific ancillary software details with version numbers. |
| Experiment Setup | Yes | Regularization parameters (for regularized logis tic regression in our algorithm, EFP, and BR; and for the margin-based objective in LIMO) were chosen by 5-fold cross-validation on the training set from {10 4 , . . . , 103} (for all algorithms, the parameter value maximizing aver age F1-measure across the 5 folds was selected). (Section 7.2) |