Original scientific paper

https://doi.org/10.17559/TV-20240614001775

Dual Optimized Event Prediction Using Meta-Heuristic Algorithms with a Distributed Deep Model for Multi-Event Forecasting

K. Uma Maheswari orcid.org/0000-0001-7302-7317 ; Department of Information Technology, University College of Engineering, BIT Campus, Anna University, Tiruchirappalli, India *
Haya Mesfer Alshahrani ; Department of Information Systems, College of Computer and Information Sciences, Princess Nourah Bint Abdulrahman University, P. O. Box 84428, Riyadh 11671, Saudi Arabia
Faiz Abdullah Alotaibi ; Department of Information Science, College of Humanities and Social Sciences, King Saud University, P. O. Box 28095, Riyadh 11437, Saudi Arabia
Prasanna Kulkarni ; Symbiosis Institute of Digital and Telecom Management (SIDTM), Symbiosis International (Deemed University), Pune, India

* Corresponding author.

Abstract

In today's technology-driven decision-making era, there is a growing demand for methodical solutions that leverage hybrid nature-inspired protocols with deep learning models (DLM). An evolution in data management through structured systems is essential. We propose a revolutionary method that combines two algorithms with distributed learning to address vast and high-velocity data streams, tackling challenges associated with noisy and imbalanced raw data sources. This study introduces an innovative integration of two nature-inspired protocols for feature selection, specifically targeting multi-event and unbalanced medical data sources in time-series-based event prediction models for source optimization. Our work utilizes the Dragonfly and Tuna Swarm algorithms within a hybrid optimization framework for feature selection. Additionally, we designed a Distributed Deep Model (DDM) to achieve high classification and prediction accuracy for multi-event data sources. Our proposed Dual Optimized Event Prediction with Distributed Deep Model (Dual-OEP-DDM) excels across key performance metrics, including accuracy, precision, recall, and F1-score. Comprehensive evaluation in dynamic event environments demonstrates that our model achieves an accuracy of 99.94%, sensitivity of 99.86%, F1-score of 99.87% (in single-event scenarios), and specificity of 99.89%, showcasing its superiority over existing models.

Keywords

data classification; distributed deep learning; hybrid optimization; nature-inspired algorithms; multi-event prediction

Hrčak ID:

332811

URI

https://hrcak.srce.hr/332811

Publication date:

29.6.2025.

Visits: 430 *