Evaluate then Cooperate: Shapley-based View Cooperation Enhancement for Multi-view Clustering

Authors: Fangdi Wang, Jiaqi Jin, Jingtao Hu, Suyuan Liu, Xihong Yang, Siwei Wang, Xinwang Liu, En Zhu

NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Comprehensive experimental results well support the effectiveness of our framework adopting to existing DMVC frameworks, demonstrating the importance and necessity of enhancing the cooperation among views.
Researcher Affiliation Academia 1 National University of Defence Technology 2 Academy of Military Science {wangfangdi19, wangsiwei13, xinwangliu, enzhu}@nudt.edu.cn
Pseudocode Yes Algorithm 1: Shapley-based Cooperation Enhancing Multi-view Clustering(SCE-MVC)
Open Source Code Yes Question: Does the paper provide open access to the data and code, with sufficient instructions to faithfully reproduce the main experimental results, as described in supplemental material? Answer: [Yes] Justification: The code has been open access.
Open Datasets Yes Our experiments utilized six multi-view datasets, including CUB2, Caltech101-73, Hand Written4, UCI-digit5, STL106 and Reuters7. The detailed information of these datasets is listed in the Table 1. CUB2: http://www.vision.caltech.edu/visipedia/CUB-200.html Caltech101-73: https://data.caltech.edu/records/mzrjq-6wc02 Hand Written4: https://archive.ics.uci.edu/ml/datasets/Multiple+Features UCI-digit5: https://cs.nyu.edu/~roweis/data.html STL106: https://cs.stanford.edu/~acoates/stl10/ Reuters7: http://archive.ics.uci.edu/ml/datasets/Reuters-21578+Text+Categorization+Collection. We conducted experiments on three additional datasets with two views: Wikipedia Articles[54], SCENE15[55], and Noisy MNIST8. Noisy MNIST8: http://yann.lecun.com/exdb/mnist/
Dataset Splits No The paper does not explicitly provide training/test/validation dataset splits. While it lists benchmark datasets, it does not specify the percentages, counts, or predefined validation sets used for its experiments.
Hardware Specification Yes We conduct all experiments on Py Torch platform using the NVIDIA 2060 GPU. Considering the scale of the Noisy MNIST dataset, all experiments were conducted on Py Torch platform using the NVIDIA 3090 GPU.
Software Dependencies No The paper mentions using 'Py Torch platform' but does not specify its version number or list any other software dependencies with their specific version numbers required for replication.
Experiment Setup Yes In sensitive analysis, we varied τ within {0.5, 1, 1.5, 2, 2.5, 3}, as illustrated in Table 4. This specifies a key hyperparameter and the values used in the experimental setup.