Feature Selection for High-Dimensional and Imbalanced Biomedical Data Based on Robust Correlation Based Redundancy...

Area of Science:

• Machine Learning
• Data Science
• Bioinformatics

Background:

• Training machine learning algorithms on imbalanced datasets with high dimensionality presents significant challenges.
• These challenges are particularly pronounced in real-world applications like biomedical data analysis.
• Existing methods often address either imbalanced data or high dimensionality, but rarely both simultaneously.

Purpose of the Study:

• To propose a novel feature selection method to address the combined problem of high dimensionality and imbalanced classes.
• To enhance classification performance for datasets with limited samples and numerous features.
• To develop an effective approach for selecting discriminative features representing both minority and majority classes.

Main Methods:

• A new method, Robust Correlation Based Redundancy and Binary Grasshopper Optimization Algorithm (rCBR-BGOA), was developed.
• rCBR-BGOA utilizes an ensemble of multi-filters combined with the Correlation-Based Redundancy method for optimal feature subset selection.
• A binary Grasshopper Optimization Algorithm (BGOA) was employed to frame feature selection as an optimization problem.

Main Results:

• The proposed rCBR-BGOA method demonstrated improved classification performance.
• Improvements were observed in key metrics such as G-mean and Area Under the Curve (AUC).
• Statistical analysis supported the effectiveness of rCBR-BGOA on high-dimensional and imbalanced datasets.

Conclusions:

• The rCBR-BGOA method effectively tackles the challenges of high-dimensional and imbalanced datasets.
• This approach offers a promising solution for improving machine learning model performance in complex data scenarios.
• The study highlights the potential of integrating feature selection with optimization algorithms for enhanced classification.