Accelerating the Big Data Analytics by GPU-Based Machine Learning: A Survey

Abstract

Today a large volume of structured and unstructured data is being generated online; the main sources for big data are social media profiles, MOOC (massive open online courses) log, social influencer, Internet of Things (IoT) data, the web, transactional applications, stream monitoring technologies, NoSQL (not only structured query language) stored data, log files, legacy document, and so on. There is a need to analyze such huge volume of data at a faster rate by uncovering the hidden patterns and correlation between the data to provide intelligent business decisions with high accuracy. The GPU (graphics processing unit)-enabled machine learning-based techniques are the strongest solution being used to perform big data analytics operation at an accelerated speed. This paper discusses selective GPU-based machine learning algorithms like decision tree, neural network, random forest, Q-learning, SARSA learning, K-means, NB (naive Bayes), AdaBoost, deep learning, support vector machine (SVM), linear regression, logistic regression, Apriori, and HMM (hidden Markov model) being used for big data analysis.