ABSTRACT
To suppress the pitfalls of neural networks, a recent learning algorithm, extreme learning machine (ELM), is proposed. ELMs minimize human intervention at the time of initialization of the disparate parameters, which aids in faster performance. ELMs not only provide a least generalization presentation at tremendously swift learning speed but also it avoid some trouble arising in gradient-based learning algorithms. There is some metamorphosis of ELMs needed to overcome some of the limitations of the method. Over time, different variants of ELMs have been developed to overcome different issues related to these limits. ELMs have been used to solve an array of pattern recognition problems including audio signal processing, handwriting analysis, and regression analysis, to name a few. In this chapter, a brief discussion about the standard ELM is presented followed by its variants and its applications in disparate fields.
