Extreme Learning Machine for land cover classification

TL;DR

This paper evaluates the effectiveness of extreme learning machine (ELM) for land cover classification using remote sensing data, demonstrating comparable accuracy to neural networks but with significantly lower computational cost.

Contribution

It introduces the application of ELM to land cover classification and compares its performance with backpropagation neural networks.

Findings

ELM achieves similar classification accuracy to neural networks.

ELM has much lower computational cost.

ELM requires only one parameter to set.

Abstract

This paper explores the potential of extreme learning machine based supervised classification algorithm for land cover classification. In comparison to a backpropagation neural network, which requires setting of several user-defined parameters and may produce local minima, extreme learning machine require setting of one parameter and produce a unique solution. ETM+ multispectral data set (England) was used to judge the suitability of extreme learning machine for remote sensing classifications. A back propagation neural network was used to compare its performance in term of classification accuracy and computational cost. Results suggest that the extreme learning machine perform equally well to back propagation neural network in term of classification accuracy with this data set. The computational cost using extreme learning machine is very small in comparison to back propagation neural network.