Big Data Challenges and Learning Paradigms: A Review

Abstract

Big data is one of the impacts of information revolution due to technological advancements such as communication, mobile and cloud services. The uncontrolled accumulation of structured and unstructured enormous volumes of data creates challenges in storing and manipulating data and obtaining valuable insights from these data. Big Data Analytics is progressively becoming popular and the organizations are in forefront to devise and adopt diversified approaches including machine learning for Big Data Analytics. Business organizations are using data learning as a scientific method for dealing with big data. The use of appropriate data analytics tools is crucial for the organizations to withstand in their business, to face the challenges in the market and gain out of competitive advantage. By considering the overwhelming demand on the data analytics tools, this review paper presents the comprehensive view on various Big Data Analytics methods in place and the state-of-the-art approaches towards Big Data Analytics. This paper also presents upcoming challenges towards big data and suggests certain mechanisms to thwart those challenges.

Jaccard Index Cat Gradient Boosting Classification for Secured Big Data Communication

“…In this section, the experimental evaluation of the proposed Jaccard Index Cat Gradient Boosting Classification based Secured Data Communication (JICGBC-SDC) method and the existing Energy-efficient Big Data-based Secure (EBDS) [1] method, Incentive-based Protection and Recovery Strategy (Incentive-based PRS) [2] is implemented in the Java language via CloudSim simulation. In order to conduct the experiment and perform secure big data communication between the cloud users, the Mobile Health dataset is used.…”

“…The quantitative performance evaluation of the JICGBC-SDC method and the existing EBDS [1] and incentive-based PRS [2] are compared with certain parameters such as classification accuracy, classification time, and error rate with respect to the number of users. The performance of proposed and existing methods is discussed with the help of a table and graphical representation.…”

“…With this, the accuracy rate of classification using the three methods, JICGBC-SDC, [1] and [2], was observed to be 98.455, 97.32%, and 96.53%, respectively. With this, the classification accuracy using JICGBC-SDC was found to be better than [1] and [2]. The reason behind the improvement was the application of the Cat Gradient Boosting classifier that first obtained the weak class learners using the Jaccard Similarity Index.…”

“…From the figure, it is inferred that increasing the frequency of cloud users results in a small decrease in the classification accuracy performance. However, with simulations conducted for 10,000 users, 98455 cloud users were properly found to be authorized as authorized using JICGBC-SDC, 97325 cloud users were properly found to be authorized as authorized using [1], and 96535 cloud users were properly found to be authorized as authorized using [2]. With this, the accuracy rate of classification using the three methods, JICGBC-SDC, [1] and [2], was observed to be 98.455, 97.32%, and 96.53%, respectively.…”

“…However, with simulations conducted for 10,000 users, 98455 cloud users were properly found to be authorized as authorized using JICGBC-SDC, 97325 cloud users were properly found to be authorized as authorized using [1], and 96535 cloud users were properly found to be authorized as authorized using [2]. With this, the accuracy rate of classification using the three methods, JICGBC-SDC, [1] and [2], was observed to be 98.455, 97.32%, and 96.53%, respectively. With this, the classification accuracy using JICGBC-SDC was found to be better than [1] and [2].…”