Incremental Nyström-based Multiple Kernel Clustering

Authors: Yu Feng, Weixuan Liang, Xinhang Wan, Jiyuan Liu, Suyuan Liu, Qian Qu, Renxiang Guan, Huiying Xu, Xinwang Liu

AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments on the proposed INMKC demonstrate its effectiveness and efficiency compared to state-of-the-art methods. In this section, we conduct experiments to verify the effectiveness and efficiency of INMKC. Specifically, the clustering performance, running time, kernel-fusion performance, convergence, and ablation study. Table 2 presents the clustering results on eight benchmark datasets.
Researcher Affiliation Academia 1College of Computer Science and Technology, National University of Defense Technology, Changsha, China, 410073 2College of Systems Engineering, National University of Defense Technology, Changsha, China, 410073 3School of Computer Science and Technology, Zhejiang Normal University, Jinhua, China, 321004 EMAIL
Pseudocode Yes Algorithm 1: Incremental Nystr om-based Multiple Kernel Clustering
Open Source Code No The implementation codes of the aforementioned methods are publicly available in their respective papers, and we run them without any modifications. This sentence refers to the baseline methods and does not explicitly state that the authors' own code for INMKC is publicly available or provide a link.
Open Datasets Yes Table 1 lists eight widely used public MVC datasets, including CUB1, PFold2, Fashion (Xiao, Rasul, and Vollgraf 2017), ALOI3, YTF104, NMNIST5, YTF1006, Winnipeg7. 1http://www.vision.caltech.edu/visipedia/CUB-200.html 2mkl.ucsd.edu/dataset/protein-fold-prediction 3https://www.kaggle.com/alvations/aloi-dataset 4https://www.cs.tau.ac.il/ wolf/ytfaces/ 5http://yann.lecun.com/exdb/mnist/ 6https://www.cs.tau.ac.il/ wolf/ytfaces/ 7https://archive.ics.uci.edu/ml/datasets/Crop+mapping+using+ fused+optical-radar+data+set
Dataset Splits No The paper does not provide specific train/test/validation dataset splits. It mentions using k-means 50 times to reduce randomness for clustering assignments, but not how the datasets themselves were partitioned for evaluation beyond being benchmark datasets.
Hardware Specification Yes All experiments are conducted on a computer equipped with an Intel Core i9-9900K CPU and 48GB RAM.
Software Dependencies No The paper mentions that 'The implementation codes of the aforementioned methods are publicly available in their respective papers', referring to baseline methods. However, it does not specify any software dependencies (e.g., libraries, frameworks) with version numbers used for their own proposed method (INMKC).
Experiment Setup Yes In our experiment, the dimension of the Nystr om approximation matrix is set to s = 300. Considering that all algorithms eventually require performing k-means to obtain final clustering assignments, we repeat k-means 50 times to reduce the randomness and report the maximum values in Table 2. We employ three evaluation metrics, including Accuracy (ACC), Normalized Mutual Information (NMI), and Purity, to verify the clustering performance.