Fast Unpaired Multi-view Clustering
Authors: Xingfeng Li, Yuangang Pan, Yinghui Sun, Quansen Sun, Ivor Tsang, Zhenwen Ren
IJCAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The efficacy, efficiency, and superiority of our FUMC are corroborated through extensive evaluations on numerous benchmark datasets with shallow and deep SOTA methods. |
| Researcher Affiliation | Academia | 1Department of Computer Science, Nanjing University of Science and Technology 2Centre for Frontier AI Research, Agency for Science, Technology and Research, Singapore 3Institute of High Performance Computing, Agency for Science, Technology and Research, Singapore 4School of Computer Science and Engineering, Southeast University 5School of National Defence Science and Technology, Southwest University of Science and Technology |
| Pseudocode | Yes | Algorithm 1 The algorithm of FUMC |
| Open Source Code | No | The paper does not provide an explicit statement that the source code for FUMC is open-source or publicly available, nor does it provide a link to such code. It only mentions collecting code for competitors or requesting it from authors. |
| Open Datasets | Yes | In the experiments, seven fully paired multi-view data are employed to compare with shallow methods, including ORL, BBCSport, MSRCv1, Mnist, YTF-10, YTF-20, and YTF-100 [Chen et al., 2022; Lin et al., 2022]. Four partially paired multi-view data with pair ratio (ϕ = 50%) to compare with deep methods, including Scene-15, Animal, Land Use-21, and RGBD [Yang et al., 2023]. Their detailed statistical information is summarized in Table 1. |
| Dataset Splits | No | The paper mentions 'partially paired multi-view data with pair ratio (ϕ = 50%)' but does not provide explicit training, validation, or test dataset split percentages, counts, or specific predefined split citations for reproducibility. |
| Hardware Specification | Yes | For shallow methods, our experiments are conducted on a 32GB RAM and Intel Core i7 CPU, 2021 Mac mini computing platform with Matlab 2021b, while deep methods are on Py Torch 1.8.1, and an NVIDIA 3090 GPU with Ubuntu 18.04. |
| Software Dependencies | Yes | Matlab 2021b... Py Torch 1.8.1, and an NVIDIA 3090 GPU with Ubuntu 18.04. |
| Experiment Setup | Yes | Algorithm 1 involves three parameters to be set properly, parameters α β, γ. Through this paper, only m = 2c anchors are learned and fixed for all datasets. The consensus dimension is also fixed as k = 2c... where µ = 1e-4 and µmax = 1010, and the complexity is O(n). The regularization parameter λ = 1/(max(m, v)n)^0.5. |