An Optimal Transport View for Subspace Clustering and Spectral Clustering
Authors: Yuguang Yan, Zhihao Xu, Canlin Yang, Jie Zhang, Ruichu Cai, Michael Kwok-Po Ng
AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We develop an alternating optimization algorithm to solve the resultant problems, and conduct experiments in different settings to evaluate the performance of our proposed methods. Experiments are conducted on simulation data and benchmark datasets to evaluate the performance of our proposed methods. |
| Researcher Affiliation | Academia | 1School of Computer Science, Guangdong University of Technology, Guangzhou, China 2Department of Mathematics, The University of Hong Kong, Hong Kong, China 3Peng Cheng Laboratory, Shenzhen, China 4Department of Mathematics, Hong Kong Baptist University, Hong Kong, China |
| Pseudocode | No | Due to the page limitation, we present the optimization algorithms for Problems (36) and (40) in the appendix. |
| Open Source Code | No | The paper does not provide any statements or links regarding open-source code for the methodology described. |
| Open Datasets | Yes | We conduct experiments on benchmark datasets from UCI machine learning repository (Asuncion and Newman 2007). |
| Dataset Splits | No | The paper mentions using benchmark datasets but does not provide specific details on training, validation, and test splits (percentages, sample counts, or predefined splits). |
| Hardware Specification | No | The paper does not specify any hardware details such as GPU models, CPU types, or memory used for running the experiments. |
| Software Dependencies | No | The paper does not provide specific software dependencies with version numbers. |
| Experiment Setup | Yes | Figure 3 shows the results of parameter sensitivity on the iris data. We observe that the performance of SFGWH is relatively stable with respect to λH, and λX has a larger effect on the performance. Similar observations can be drawn from the other datasets. ... In specific, we calculate CX and CH based on the squared Euclidean distance defined in Eq. (3), and then construct CX,H as CX,H = CX + λCH, where λ control the effects of CX and CH. ... where λX = α, λH = αλ, and ϵ = 1 in our experiments. ... Following the setting used in (Liu et al. 2019; Zhang et al. 2021), we take the samples from the smallest clusters as outliers. ... We follow (Liu et al. 2019) to set the number of clusters as the true number plus one for K-means... |