Extreme Learning Machine

Define classification problems in 2-D space and solve with an ELM Neural Network

This app solves dual-class classification problems in a 2-dimensional space with an Extreme Learning Machine (ELM) Artificial Neural Network (ANN). The algorithm used is based on the paper “Extreme Learning Machine: Theory and Applications” by G.-B. Huang, Q.-Y. Zhu and C.-K. Siew, Neurocomputing, vol. 70, pp. 489-501, 2006.

It is designed for educational use and may be useful when teaching courses in Machine Learning, whilst also demonstrating the efficiency of the ELM as a classification tool.

Instructions for use:

First, define a dual class problem using the red and blue paints by selecting "Define Problem" in the upper left corner of the screen and drawing on the canvas.

Press "Training Data" and drag the slider to define how many classified points are used to train the network.

Press "Setup Network" to specify the number of hidden layer neurons to use.

Press "Train" to train the network - this may take between 5-30 seconds depending on your hardware and problem setup.

On the top right of the screen, once the network is trained, you can switch between viewing the original specified problem, and the canvas space as defined by your trained neural network.

For more information on ELMs, references and codes for more complex problems visit www3.ntu.edu.sg/home/egbhuang/