Learning ELM network weights using linear discriminant analysis

TL;DR

This paper introduces a linear discriminant analysis-based method for determining weights in Extreme Learning Machines, offering an alternative to the pseudo-inverse approach with potentially optimal classification performance.

Contribution

It proposes a novel linear discriminant analysis method for ELM weight calculation, improving upon existing pseudo-inverse techniques.

Findings

Provides Bayes optimal weight estimates for ELMs.

Offers an alternative method that may enhance classification accuracy.

Simplifies weight determination process in ELMs.

Abstract

We present an alternative to the pseudo-inverse method for determining the hidden to output weight values for Extreme Learning Machines performing classification tasks. The method is based on linear discriminant analysis and provides Bayes optimal single point estimates for the weight values.