Support matrix machine: A review

Area of Science:

• Machine Learning
• Data Science
• Pattern Recognition

Background:

• Support Vector Machine (SVM) is a popular machine learning algorithm for classification and regression.
• SVM requires vectorized data, necessitating reshaping of matrix data, which disrupts spatial correlations and increases dimensionality.
• High dimensionality in vectorized matrix data leads to significant computational complexity.

Purpose of the Study:

• To introduce and analyze Support Matrix Machine (SMM) as a novel methodology for classifying matrix input data.
• To highlight SMM's ability to overcome the limitations of traditional SVM when dealing with matrix-formatted data.
• To provide a comprehensive overview of SMM development for researchers and practitioners.

Main Methods:

• Support Matrix Machine (SMM) is proposed to handle matrix data directly.
• SMM utilizes the spectral elastic net property, combining nuclear norm and Frobenius norm, to preserve matrix structure.
• Analysis covers various SMM variants including robust, sparse, class-imbalance, and multi-class classification models.

Main Results:

• SMM effectively classifies matrix data by preserving inherent structural information.
• The spectral elastic net property ensures efficient handling of matrix data without loss of spatial correlations.
• SMM variants offer tailored solutions for diverse classification challenges with matrix data.

Conclusions:

• SMM is a promising approach for machine learning tasks involving matrix data.
• The preservation of structural information and reduced computational complexity make SMM advantageous over traditional SVM for matrix data.
• Future research directions are identified to further advance SMM algorithms and applications.