Revisiting Pseudo-Label for Single-Positive Multi-Label Learning

Authors: Biao Liu, Ning Xu, Jiaqi Lv, Xin Geng

ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on four image datasets and five MLL datasets show the effectiveness of our methods over several existing SPMLL approaches.
Researcher Affiliation Academia 1School of Computer Science and Engineering, Southeast University, Nanjing 210096, China 2RIKEN Center for Advanced Intelligence Project, Tokyo 103-0027, Japan. Correspondence to: Ning Xu <xning@seu.edu.cn>, Xin Geng <xgeng@seu.edu.cn>.
Pseudocode Yes Algorithm 1 MIME Algorithm
Open Source Code No No explicit statement about providing open-source code or a link to a code repository was found in the paper.
Open Datasets Yes In the experiments, following (Cole et al., 2021; Xu et al., 2022), we employed four large scale multi-label image classification (MLIC) datasets and five widely-used MLL datasets (Hang & Zhang, 2022) to evaluate our proposed method. The four MLIC datasets include PSACAL VOC 2021 (VOC) (Everingham et al., 2010), MS-COCO 2014 (COCO) (Lin et al., 2014), NUS-WIDE (NUS) (Chua et al., 2009), and CUB-200 2011 (CUB) (Wah et al., 2011); the five MLL datasets cover a wide range of scenarios with heterogeneous multi-label characteristics.
Dataset Splits Yes For each MLIC dataset, we withhold 20% of the training set for validation. For each MLL dataset, we split the dataset as train/validation/test set in a ratio of 80%/10%10%.
Hardware Specification No No specific hardware details (e.g., GPU/CPU models, memory amounts, or detailed computer specifications) used for running experiments were mentioned.
Software Dependencies No The paper mentions "We use the Adam optimizer (Kingma & Ba, 2015)" but does not provide specific version numbers for software components or libraries.
Experiment Setup Yes The batch size is selected from {8, 16} and the number of epochs is set to 10. The learning rate , weight decay, and the tradeoff parameter β are selected from {10-2, 10-3, 10-4} with a validation set.