FOA-ELM: Fruit Fly Optimization Algorithm for Extreme Learning Machine

This documentation presents an innovative algorithmic approach that enhances learning efficiency through optimized neural network architecture. The methodology comprises two critical phases: 1. Determination of hidden layer neuron count in Extreme Learning Machine using Fruit Fly Optimization Algorithm. This optimization process employs FOA's search mechanism to identify the optimal number of neurons, significantly impacting network performance by balancing model complexity and generalization capability. The implementation involves configuring population parameters and fitness evaluation functions within the optimization module. 2. Training and testing of samples using Extreme Learning Machine methodology with the optimized neuron configuration. This phase leverages ELM's random projection and analytical solution characteristics, where the hidden layer weights are randomly initialized while output weights are computed through Moore-Penrose pseudoinverse, ensuring rapid learning convergence. The code structure separates data preprocessing, model training, and performance evaluation modules. To implement this algorithm, users can execute the main function in fruitfly_elm.m file, which integrates both optimization and training phases. The modular design permits straightforward dataset substitution by modifying data loading parameters, accommodating various research requirements through configurable input interfaces. The implementation includes error handling and performance metrics calculation for comprehensive algorithm validation.