Multiple Kernelized Fuzzy C-means for Heterogeneous data using Unsupervised Machine Learning

Abstract

In the context of big data, heterogeneity is a prominent characteristic, & heterogeneousdata adds to difficulties of information convergence for big data analytics. The majority of datasets are diverse in terms of shape, formation, layout, variance, punctuation, and availability, among other characteristics. The broad variety of data sources usually results in software silos, which are a collection of unintegrated data systems with diverse methods, query languages, and application programming interfaces (APIs). During pre-processing stage, it is often necessary to combine data types from disparate sources. Scalability across diverse sources of information can only be achieved by automated or minimal human interaction in holistic models of data processing. Data aggregation is a method that brings together several different local resources without storing their information in a single central repository. Kernel learning is an active study area in the ML field, with many researchers workingon it at the same time. Over the last decade, researchers have conducted substantial research into the
family of kernel-based ML techniques.ELM is a rapidly increasing-learning algorithm for a single hidden layer of feedforward neuralnetworks that may be used in both regressions as well as classification. ELMs for one hidden layer may randomly choose the node number of the hidden layer, as well as distribute input weights & hidden layer biases across the network. A mathematical modification is essential to finish the learning experience once the weights of the o/p layer have been established
using the least square technique.