Knowledge discovery employing grid scheme least squares support vector machines based on orthogonal design bee colony...

Area of Science:

• Machine Learning
• Data Mining
• Computational Intelligence

Background:

• Least Squares Support Vector Machine (LSSVM) is a powerful classification tool.
• Feature selection and parameter optimization are crucial for LSSVM performance.
• Mixed kernels can improve LSSVM's ability to handle complex data.

Purpose of the Study:

• To propose a novel machine learning paradigm, GS-LSSVM, integrating a Grid Scheme (GS) with LSSVM.
• To enable simultaneous feature selection, mixed kernel application, and parameter optimization within a unified framework.
• To enhance the interpretability and accuracy of classification models.

Main Methods:

• Integration of a Grid Scheme (GS) with LSSVM.
• Utilizing orthogonal design for initializing features and parameters in GS.
• Employing random feature selection based on the GS.
• Applying an artificial bee colony algorithm for LSSVM parameter optimization.
• Testing on ten datasets from the UCI database.

Main Results:

• GS-LSSVM produces classification models that are more easily interpreted.
• The approach effectively reduces the number of features required for classification.
• GS-LSSVM demonstrates significantly superior classification accuracy (hit ratio) compared to other methods.
• The proposed method shows promise in classification exploration.

Conclusions:

• GS-LSSVM offers an effective approach for data classification problems.
• The integration of GS enhances model interpretability and reduces feature dimensionality.
• The method achieves high accuracy, outperforming existing techniques.
• GS-LSSVM is a promising direction for future research in machine learning classification.