Optimizing Workforce Demand Forecasting in Enterprises: A Novel Approach with SF-SOA-based Extreme Learning Machine

The Extreme Learning Machine (ELM) stands out in machine learning as a powerful tool for approximating complex nonlinear mappings. However, challenges arise from the inherent randomness in weight initialization, impacting consistency and optimality. This work introduces a novel optimization approach, the Spiral Flight Seagull Optimization Algorithm-based ELM (SF-SOA-based ELM), leveraging the Seagull Optimization Algorithm (SOA) for global and local search capabilities. SF-SOA optimizes weights and thresholds, addressing traditional SOA limitations with three key strategies: chaotic sequences, optimization of spiral flight paths and local search based on convergence conditions. Applied to practical workforce demand forecasting in enterprises, SF-SOA demonstrates significant improvements in convergence accuracy, speed, and error reduction. The refined SF-SOAbased ELM showcases enhanced performance, highlighting its potential to overcome challenges in random weight initialization within traditional networks.