Tensor Multi-Clustering Parallel Intelligent Computing Method Based on Tensor Chain Decomposition.

Area of Science:

• Data Science
• Machine Learning
• Multi-view Learning

Background:

• Increasing use of multi-sensor technologies generates large, high-dimensional datasets.
• Traditional clustering and Principal Component Analysis (PCA) struggle with complex, multi-view data.
• Tensors offer a powerful way to represent multi-dimensional data, capturing richer semantic information.

Purpose of the Study:

• To investigate and evaluate advanced multi-clustering algorithms for large-scale, high-dimensional data.
• To explore tensor decomposition methods as alternatives to PCA for uncovering latent components.
• To assess the performance of these methods in dimensionality reduction and supervised learning.

Main Methods:

• In-depth performance evaluation of three multi-clustering algorithms: Self-weighted Multiview Clustering (SwMC), Latent Multi-view Subspace Clustering (LMSC), and Multi-view Subspace Clustering with Intactness-Aware Similarity (MSC IAS).
• Comparative analysis of various tensor decomposition techniques against classical PCA for extracting latent components.
• Experimental testing on seven real-world datasets using key metrics: Accuracy (ACC), normalized mutual information (NMI), and purity.

Main Results:

• The evaluated multi-clustering algorithms demonstrate effectiveness in handling complex, high-dimensional datasets.
• Tensor decomposition methods show superior capability in extracting latent components compared to traditional PCA.
• Both tensor decomposition and multi-clustering approaches show promise for dimensionality reduction and supervised learning tasks.

Conclusions:

• Advanced multi-clustering and tensor decomposition techniques are essential for analyzing modern, high-dimensional, multi-view data.
• These methods offer significant advantages over traditional approaches in terms of performance and insight extraction.
• Further research into tensor models can enhance capabilities in dimensionality reduction and supervised learning.