A review of advances in extreme learning machine techniques and its applications

Abstract

Feedforward neural networks (FFNN) has been used for machine learning researches, and it really has a wide acceptance. It was noted in the recent time that feedforward neural network is far slower than required. This has created a serious bottleneck in its applications. Extreme Learning Machines (ELM) had been proposed as alternative learning algorithm to FFNN, which is characterized by single-hidden layer feedforward neural networks (SLFN). It randomly chooses hidden nodes and determines their output weight analytically. This paper review is to provide a roadmap for ELM as an efficient research tool in machine learning with the aim of finding research gap into further study. It was discovered through this study that research publications in ELM continues to grow yearly from 16.20% in 2013 to 40.83% in 2016.