Semi-Supervised Partial Label Learning via Confidence-Rated Margin Maximization

Authors: Wei Wang, Min-Ling Zhang

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments on synthetic as well as real-world data sets clearly validate the effectiveness of the proposed semi-supervised partial label learning approach.
Researcher Affiliation Academia Wei Wang Min-Ling Zhang School of Computer Science and Engineering, Southeast University, Nanjing 210096, China Key Laboratory of Computer Network and Information Integration (Southeast University), Ministry of Education, China {wang_w, zhangml}@seu.edu.cn
Pseudocode Yes Table 1: Pseudo-code of PARM.
Open Source Code No The paper does not provide a specific link to source code or explicitly state that the code for the methodology is being released or is publicly available.
Open Datasets Yes Table 2 summarizes characteristics of the experimental data sets used in this paper. Following the widely-used experimental protocol in partial label learning [6, 7, 8, 11], synthetic PL data sets are generated from multi-class UCI data sets with controlling parameter r. Furthermore, five real-world PL data sets from different task domains have also been employed for experimental studies, including Lost [8], LYN10, LYN20 [12] for automatic face naming, Mirflickr[13] for web image classification, and Bird Song [4] for bird song classification.
Dataset Splits Yes On each data set, ten-fold cross validation is performed whose mean accuracy as well as standard deviation are recorded for all comparing approaches.
Hardware Specification No The paper does not provide specific hardware details (like CPU/GPU models or memory amounts) used for running the experiments. It only mentions: "We thank the Big Data Center of Southeast University for providing the facility support on the numerical calculations in this paper."
Software Dependencies No The paper does not provide specific software dependencies with version numbers (e.g., Python, PyTorch, specific libraries) needed to replicate the experiment.
Experiment Setup Yes In this paper, σ, k and α are fixed to be 1, 8 and 0.95 respectively. ...the regularization parameters λ and µ for PARM are chosen among {0.001, 0.005, 0.01, 0.05, 0.1, 0.5, 1, 5, 10} via cross-validation on training set and γ = 0.01.