Clustering Ensemble Meets Low-rank Tensor Approximation

Authors: Yuheng Jia, Hui Liu, Junhui Hou, Qingfu Zhang7970-7978

AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results over 7 benchmark data sets show that the proposed model achieves a breakthrough in clustering performance, compared with 11 state-of-the-art methods.
Researcher Affiliation Academia 1School of Computer Science and Engineering, Southeast University, Nanjing 210096, China 2Department of Computer Science, City University of Hong Kong, Kowloon, Hong Kong SAR
Pseudocode Yes Algorithm 1 t-SVD of a 3-D tensor (Zhang et al. 2014) and Algorithm 2 Numerical solution to Eq. (9) are provided.
Open Source Code Yes To reproduce the results, we made the code publicly available at https://github.com/jyhlearning/Tensor Clustering Ensemble.
Open Datasets Yes Following recent clustering ensemble papers (Huang, Wang, and Lai 2018; Huang, Lai, and Wang 2016; Zhou, Zheng, and Pan 2019), we adopted 7 commonly used data sets, i.e., Bin Alpha, Multiple features (MF), MNIST, Semeion, Cal Tech, Texture and ISOLET.
Dataset Splits No The paper does not explicitly provide details about train/validation/test dataset splits. It mentions randomly selecting samples and base clusterings for repetitions, but not data partitioning for validation purposes.
Hardware Specification No The paper does not specify the hardware (e.g., CPU, GPU models, memory) used for the experiments.
Software Dependencies No The paper does not provide specific version numbers for any software dependencies used in the experiments (e.g., programming languages, libraries, frameworks).
Experiment Setup Yes For the compared methods, we set the hyper-parameters according to their original papers. If there are no suggested values, we exhaustively searched the hyper-parameters, and used the ones producing the best performance. The proposed model only contains one hyperparameter λ, which was set to 0.002 for all the data sets.