A Comparative Study of Extreme Learning Machine Pruning Based on Detection of Linear Independence

Abstract

Extreme Learning Machine (ELM) is gaining fairly popularity in training neural networks, due to its simplicity and speed. However, the number of neurons in the hidden layer is still an open problem. This paper proposes a method for pruning the hidden layer neurons based on the linear combination of the hidden layer weights and the input data and compare four methods of detecting linear dependence between vectors.