ELM Resources

Showing items tagged with "ELM"

Comparative analysis of ELM (Extreme Learning Machine) and ELM_2 algorithms with source code implementations including EELM (published in Neural Computing by Yuguang Wang) and YELM (enhanced ELM version published by Yuboyuan). This program provides performance benchmarking for these three machine learning approaches. For theoretical background, please refer to the authors' publications. To execute the comparison, simply run the pendigits_mean.m main file which handles data preprocessing, model training, and performance evaluation.

MATLAB 301 views Tagged
Download

Extreme Learning Machine (ELM) is a simple, efficient learning algorithm for Single-hidden Layer Feedforward Networks (SLFNs) proposed by Associate Professor Guang-Bin Huang from Nanyang Technological University in 2006. Unlike traditional neural network training algorithms (e.g., Backpropagation), ELM requires minimal parameter tuning—only the number of hidden nodes needs specification—and avoids local optima by randomly initializing input weights and biases without iterative adjustments. The algorithm computes output weights analytically via Moore-Penrose pseudoinverse, ensuring unique optimal solutions and delivering rapid training with strong generalization. Code implementations typically involve random weight initialization, hidden layer activation (e.g., sigmoid), and linear output solving.

MATLAB 215 views Tagged
Download