Novel chaotic oppositional fruit fly optimization algorithm for feature selection applied on COVID 19 patients'...

Area of Science:

• Computer Science
• Artificial Intelligence
• Data Science

Background:

• The increasing volume of data poses challenges for machine learning, leading to high computational costs and reduced performance.
• Feature selection is crucial for optimizing datasets by identifying the most relevant features, but high-dimensional data complicates this process.
• Existing optimization algorithms often struggle to balance dimensionality reduction with accuracy preservation.

Purpose of the Study:

• To propose a novel chaotic opposition fruit fly optimization algorithm tailored for binary optimization problems.
• To enhance the efficiency and effectiveness of feature selection in high-dimensional datasets.
• To address the limitations of traditional optimization algorithms in machine learning preprocessing.

Main Methods:

• Development of a chaotic opposition fruit fly optimization algorithm, an adaptation of the original fruit fly algorithm.
• Testing the algorithm on ten unconstrained benchmark functions.
• Evaluation using twenty-one standard datasets from the UCI repository and Arizona State University, including a coronavirus disease dataset.

Main Results:

• The proposed algorithm demonstrates superior performance compared to established feature selection methods.
• It effectively reduces the number of features utilized in datasets.
• Significant improvements in classification accuracy were observed across various datasets.

Conclusions:

• The chaotic opposition fruit fly optimization algorithm is a highly effective method for feature selection.
• This approach offers a promising solution for improving machine learning model efficiency and accuracy.
• The algorithm's adaptability makes it suitable for diverse datasets, including those in medical applications.