Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Machine learning as a service for enabling Internet of Things and People
Assem H., Xu L., Buda T., O’Sullivan D.  Personal and Ubiquitous Computing 20 (6): 899-914, 2016. Type: Article
Date Reviewed: Feb 8 2017

The huge amount of data being produced and disseminated through Internet of Things (IoT) platforms every day generates demand for more complex computing solutions to handle data transfers between billions of devices connected to the Internet. The authors found a new approach to the field, transforming the concept of IoT to the Internet of People (IoP), making the Internet a platform for converged points of connection between machines, devices, or humans. By its nature, IoT combines interactions between connected devices. The process of capturing and interpreting the data that these devices produce, as well as taking action according to the results of the data processing, requires complex machine learning. Since machine learning has a broad scope, the authors focus their research on supervised learning models based on regression and classification models in which datasets are used in a learning process. The recommender system has a vital role in this process and the paper provides the unique approach that the authors call TCDC (short for “train, compare, decide, and change”); they also present machine learning as a service--an approach different from commercial machine learning as a service platforms.

TCDC is used to select the optimum machine learning model. It consists of four basic phases, mentioned above. Besides the process of finding the best acceptable predictive performance of the model, other factors such as interoperability, computational complexity, and ease of implementation are also included in order to successfully compare TCDC with the reference model. The authors explain their TCDC-based regression recommender system in detail, and help readers better understand the problem of machine learning in this context.

The evaluation section makes this work really impressive, going into a new and useful approach that provides better insight into machine learning models in IoT ecosystems--the authors measured the predictive performance of a model and presented regression and classification results. Further, it includes a brief but well-documented description of results showing the positive validation of the TCDC approach through pretty high predictive accuracy points both for regression and classification tests.

Reviewer:  F. J. Ruzic Review #: CR145046 (1705-0304)
Bookmark and Share
  Editor Recommended
Featured Reviewer
Learning (I.2.6 )
Correlation And Regression Analysis (G.3 ... )
Model Classification (I.6.1 ... )
Would you recommend this review?
Other reviews under "Learning": Date
ETARM: an efficient top-k association rule mining algorithm
Nguyen L., Vo B., Nguyen L., Fournier-Viger P., Selamat A.  Applied Intelligence 48(5): 1148-1160, 2018. Type: Article
Aug 6 2018
Multi swarm optimization algorithm with adaptive connectivity degree
Vafashoar R., Meybodi M.  Applied Intelligence 48(4): 909-941, 2018. Type: Article
Aug 6 2018
Deep learning with Python: a hands-on introduction
Ketkar N.,  Apress, New York, NY, 2017. 226 pp. Type: Book (978-1-484227-65-7)
Jul 16 2018

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright © 2000-2018 ThinkLoud, Inc.
Terms of Use
| Privacy Policy