Incomplete Multi-view Clustering via Diffusion Contrastive Generation
Authors: Yuanyang Zhang, Yijie Lin, Weiqing Yan, Li Yao, Xinhang Wan, Guangyuan Li, Chao Zhang, Guanzhou Ke, Jie Xu
AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments demonstrate that our method outperforms state-of-the-art approaches. Experiments To evaluate the effectiveness of DCG, we conducted extensive experiments to answer the following questions: (Q1) Does DCG outperform state-of-the-art IMVC methods? (Q2) Does each component of DCG contribute to the overall performance? (Q3) Do the hyperparameters affect the performance of DCG? (Q4) What is the clustering structure revealed by DCG? |
| Researcher Affiliation | Academia | 1School of Computer Science and Engineering, Southeast University, Nanjing 210096, China 2College of Computer Science, Sichuan University 3School of Computer and Control Engineering, Yantai University 4Key Laboratory of New Generation Artificial Intelligence Technology and Its Interdisciplinary Applications (Southeast University), Ministry of Education, China 5College of Computer, National University of Defense Technology 6College of Computer Science and Technology, Zhejiang University 7Department of Control Science and Intelligence Engineering, Nanjing University 8School of Economics and Management, Beijing Jiaotong University 9School of Computer Science and Engineering, University of Electronic Science and Technology of China EMAIL, EMAIL |
| Pseudocode | No | The paper describes methods through detailed prose and mathematical equations but does not present any clearly labeled pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not contain an explicit statement about releasing source code for the described methodology, nor does it provide a link to a code repository. |
| Open Datasets | Yes | We conducted experiments on five multi-view datasets, including Synthetic3D (Kumar, Rai, and Daume 2011), CUB (Wah et al. 2011), Hand Written (Asuncion and Newman 2007), Land Use-21 (Yang and Newsam 2010), and Fashion (Xiao, Rasul, and Vollgraf 2017). |
| Dataset Splits | No | To evaluate the performance of handling incomplete multi-view data, following (Lin et al. 2021), we randomly select m instances and randomly delete one view, where the missing rate is m/n, with n representing the total number of instances. Detailed descriptions of the datasets and implementation are in the supplementary materials. |
| Hardware Specification | No | The paper does not provide specific hardware details (such as GPU models, CPU types, or memory specifications) used for running the experiments. |
| Software Dependencies | No | The paper does not provide specific ancillary software details, such as library names with version numbers, that were used to implement the experiments. |
| Experiment Setup | Yes | In our experiments, these three trade-off coefficients λ1, λ2, and λ3 are all set to 1. The total number of time steps T varies from 10 to 100. Our model is an end-to-end clustering method that does not require k-means (Bauckhage 2015) clustering to obtain the final clustering results. Therefore, we can optimize the entire model simultaneously. |