A Survey on Forecasting Methods by using Various Types of Extreme Learning Machine

Abstract

A Single Layer Feed Forward Neural webs which is a narrative learning of step by step instruction for (SLFNs) i.e., called as Extreme Learning Machine which helps in training limits of hidden nodes and also involves weights to input and the biases, that is to be set dynamically and need not require to adjust it, at the period of time the resultant weights can scientifically resolved by the operation of simple generalized inverse and exclusively one limit is applied i.e. the number of nodes which is hidden needs to explain. In contrast with other learning step by step instruction of single layer which is hidden i.e., design for feed forward neural webs, ELM gives terribly faster learning rate along with effective generalization performance. Initially it gives a short idea of ELM, explaining the basic concept and its step by step instruction. Further it focuses on the various types of ELM, mainly on incremental, two-stage, error-minimized, pruning, voting-based, ordinal, online sequential, evolutionary and fully complex ELM.