Fractal feature selection model for enhancing high-dimensional biological problems.

Area of Science:

• Bioinformatics
• Computational Biology
• Machine Learning

Background:

• Bioinformatics generates vast, complex datasets, posing challenges for accurate classification.
• Intelligent systems require effective feature selection to improve machine learning performance in bioinformatics.

Purpose of the Study:

• To introduce a novel feature selection model, Fractal Feature Selection (FFS), for high-dimensional bioinformatics problems.
• To enhance the accuracy of intelligent classification systems in bioinformatics.

Main Methods:

• The proposed FFS model utilizes a fractal concept to select relevant features.
• Features are segmented into blocks, and similarity is assessed using Root Mean Square Error (RMSE).
• Feature importance is determined by low RMSE values, indicating high relevance.

Main Results:

• The FFS model was evaluated on ten high-dimensional bioinformatics datasets.
• FFS significantly improved machine learning accuracy for classification tasks.
• Accuracy increased from an average of 79% with full features to 94% using FFS.

Conclusions:

• The Fractal Feature Selection (FFS) model offers a robust approach to feature selection in bioinformatics.
• FFS demonstrably enhances machine learning accuracy for classifying complex biological data.
• This method addresses the challenge of high-dimensional data in bioinformatics classification.