Efficient correntropy-based multi-view clustering with anchor graph embedding.

Area of Science:

• Machine Learning
• Data Mining
• Computer Science

Background:

• Multi-view clustering offers superior performance over single-view methods but suffers from high computational costs and sensitivity to noise and outliers.
• Improving the efficiency and robustness of multi-view clustering algorithms remains a significant challenge in data analysis.

Purpose of the Study:

• To propose an efficient correntropy-based multi-view clustering (ECMC) algorithm.
• To address the limitations of high computational cost and lack of robustness in existing multi-view clustering techniques.

Main Methods:

• ECMC utilizes an embedded anchor graph and nonnegative matrix factorization (NMF) to enhance clustering efficiency.
• Correntropy is employed to suppress noise and outliers, thereby improving robustness.
• A half-quadratic-based strategy optimizes the non-convex objective function, and NMF is modified using an indicator matrix for direct category acquisition.

Main Results:

• ECMC demonstrated significantly faster processing speeds compared to state-of-the-art algorithms.
• The algorithm exhibited superior robustness, effectively handling complex noises and outliers in datasets.
• Experiments on eight real-world and eighteen noisy datasets validated the performance of ECMC.

Conclusions:

• The proposed ECMC algorithm effectively balances efficiency and robustness in multi-view clustering.
• ECMC offers a promising solution for analyzing complex, noisy, real-world multi-view data.
• The novel approach of integrating correntropy and modified NMF provides a significant advancement in the field.