A Hybrid BSA-PSO Algorithm with Dynamic Parameters for Heterogeneous Data Classification

Abstract

The explosive growth and immense diversity of modern data,that often described as heterogeneous,which present a significant challenge for traditional sorting and classification methods. Handling mixe the data types like text, images, and numerical values efficiently has become a major hurdle. This research is confronts this problem by proposing the development and improvement of the Birds Algorithm (BSA),a nature-inspired optimization technique. The primary goal was to significantly enhanced both the classification accuracy and the performance speed when dealing with such complex datasets. To achieve that, it modified the algorithm collective search mechanism by introducing new dynamic parameters, that moving beyond static values. That included dynamically adjusting the inertia itight and the cognitive and social learning factors,which allowing the algorithm to intelligently adapt its search strategy over time. That hybrid approach was then rigorously tested using the heterogeneous dataset and its performance was benchmarked against to other standard artificial intelligence algorithms that including Particle Swarm Optimization (PSO) and Genatic Algorithm (GA). The results from that comparative study itre definitive,that demonstrating the clear superiority of our developed algorithm in the both accuracy and execution time. The proposed modified BSA-PSO achieved the best performance with an average execution time of 0.95 seconds and an average final fitness of 0.05, significantly outperforming the Standard PSO, which recorded an execution time of 2.45 seconds and a fitness of 4.20, as itll as the GA, which shoitd an execution time of 3.10 seconds and a fitness of 3.80. This work underscores the urgent necessity of developing such adaptive, intelligent systems and confirms that combining optimization algorithms is a highly promising path toward managing the data challenges of our modern digital world.