Golden eagle based improved Att-BiLSTM model for big data classification with hybrid feature extraction and feature...

Area of Science:

• Computer Science
• Data Science
• Artificial Intelligence

Background:

• Technological advancements have led to massive big data growth.
• Machine learning (ML) is crucial for examining and classifying big data.
• Effective feature extraction and selection are vital for ML model performance.

Purpose of the Study:

• To develop an optimized deep learning classifier for big data classification.
• To integrate hybrid feature extraction and selection methods for enhanced performance.
• To improve the accuracy and efficiency of big data analysis.

Main Methods:

• Utilized local linear embedding-based kernel principal component analysis for feature extraction.
• Employed perturbation theory with heuristic search for feature selection, optimized by five algorithms.
• Applied an attention-based bidirectional long short-term memory classifier optimized with a golden eagle-inspired algorithm.

Main Results:

• The proposed framework achieved over 90% accuracy in classifying large datasets.
• Experimental verification on publicly accessible datasets confirmed the model's effectiveness.
• The hybrid feature engineering approach significantly improved classification performance.

Conclusions:

• The developed deep learning approach offers a robust solution for big data classification.
• The integration of advanced feature engineering techniques enhances model accuracy.
• This framework provides a valuable tool for analyzing and categorizing large-scale datasets.