Maximin Separation Probability Clustering
Authors: Gao Huang, Jianwen Zhang, Shiji Song, Zheng Chen
AAAI 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Empirical results on real world data sets verify that MSP is a robust and effective clustering quality measure. It is also shown that the proposed algorithms compare favorably to state-of-the-art clustering algorithms in both accuracy and efficiency. |
| Researcher Affiliation | Collaboration | Tsinghua National Laboratory for Information Science and Technology (TNList) Department of Automation, Tsinghua University, Beijing 100084, China Microsoft Research, Beijing |
| Pseudocode | Yes | Table 1: Summary of the proposed three MSPC algorithms. This table provides structured algorithmic steps for MSPCMPM, MSPCGEP, and MSPCEIG. |
| Open Source Code | No | The paper does not contain any statements or links indicating that the source code for the described methodology is publicly available. |
| Open Datasets | Yes | All of data sets are obtained from the UCI repository, except the handwritten digits data sets MNIST4 and USPS. For those data sets originally contain more than two classes, we select their first two classes to create binary clustering tasks, if not explicitly stated. ... 4Available at http://yann.lecun.com/exdb/mnist/ |
| Dataset Splits | No | The paper mentions using UCI, MNIST, and USPS datasets for 'binary clustering tasks' and normalizes features, but it does not provide specific train/validation/test splits (e.g., percentages or sample counts) for reproducibility. |
| Hardware Specification | Yes | All algorithms are implemented in MATLABTM, and are executed on an Intel i7 Quad Core CPU 3.39GHz machine with 16GB RAM. |
| Software Dependencies | No | The paper states that 'All algorithms are implemented in MATLABTM', but it does not provide specific version numbers for MATLAB or any other software libraries or dependencies used, which is required for reproducible software description. |
| Experiment Setup | Yes | For Iter SVR, LGMMC and MVC, both the regularization parameter C and the Gaussian kernel parameter σ are tuned (linear kernel LGMMC is used for the three mnist data sets), following the settings in (Li et al. 2009) and (Niu et al. 2013), i.e., the hyperparameters correspond to highest clustering accuracy are directly selected. For LDA-KM, Dis-KM and the proposed three MSPC algorithms, the regularization coefficient, which is the only hyperparameter, is optimally chosen from the candidate set [10 4, 10 3, . . . , 104] based on the clustering accuracy. |