Machine Learning in IoT: Challenges and Opportunities

Desklib provides solved assignments and past papers to help students succeed.

Machine Learning in IOT

Problem Statement
Internet of Things is expected to be among the key increment of the latest big trends that are
taking place in the new technologies. By having the collection along the present as well as the
estimated ambiguity and the persuasiveness of the cabled smart systems it provides a base to the
applications in various areas such as the technologies like smart city, smart home, healthcare,
transportations, organizational automations and most importantly retailing . The technical
creations have allowed the internet of things related technology to decrease the price, increase
the calculating ability of the system, cloud service, interoperability and safety and privacies
enhancement. One of the key characteristics that are responsible for the success of internet of
things is the technology of Machine Learning. Machine Learning is among the features of IoT
the each and every system that has been associated with Internet of Things, must be having. But
as there have been some advancements in the technology of IoT as well as Machine Learning
there have been many problems that are observed in associating these 2 technologies together. In
the document that has been created, the dos and don’ts of internet of things are discussed. The
topic that has been chosen in for document is machine learning in IOT. In the course of this
document a learner would be having the ability to identify and resolve the problems that occur in
the implementation of AI in IoT. Some of the problems that would be discussed in the given
document are Data Consideration, Semi- Supervised Machine Learning, Crowd Sense (Predky,
2018).

Introduction
In the case of Internet of Things, with the help of obtaining the collection along the present as
well as the estimated ambiguity and the persuasiveness of the cabled smart systems it provides a
base to the applications in various areas such as the technologies like smart city, smart home,
healthcare, transportations, organizational automations and most importantly retailing. Internet
of Things consists of a very huge space in any memory as a lot of data has to be stored in order
to make it work and when it is associated with a technology such as artificial intelligence or
machine learning which consume a very large space themselves. Crowd sensing is one of the
most important aspects in the process of IoT when it comes to the implementation of Machine
Learning in IoT. The technical creations have allowed the internet of things related technology to
decrease the price, increase the calculating ability of the system, cloud service, interoperability
and safety and privacies enhancement. One of the key characteristics that are responsible for the
success of internet of things is the technology of Machine Learning. Machine Learning is among
the features of IoT the each and every system that has been associated with Internet of Things,
must be having. But as there have been some advancements in the technology of IoT as well as
Machine Learning there have been many problems that are observed in associating these 2
technologies together. The slotions for the problems that are mentioned in the problem statement
are also provided below in order to create a basic knowledge to the user when they will be
having the access to this document and finding for the solution of the problems that they are
facing in the implementation of Maching Leaning in Internet of Things
Consideration of Data: It is not an unknown fact that Internet of Things consists of a very huge
space in any memory as a lot of data has to be stored in order to make it work and when it is
associated with a technology such as artificial intelligence or machine learning which consume a
very large space themselves (Jagtaap, 2018). The space complexity of the whole system becomes
extremely high. This may cause the small problems and errors such as data redundancy, data
complexity, memory loss, etc. One of the major causes of this problem is that due to the high
amount of data without a proper management, the system becomes a bit noisy and error starts to
occur and the acquisition end as well as the transmission end (Predky, 2018). The IoT systems
are already having inconsistent data flow and with a large amount of data which might cause data
redundancy, this inconsistency is increased even more. In order to solve this problem the only
appropriate way is to store and manage the whole data properly. If the data is managed properly
them in that case there would be least problems because decreasing the use of data in IoT and
Machine Learning cannot be an option ever (Jagtaap, 2018).
Semi Supervised Machine Leaning: The data that is known to be generated in the IoT systems
comes or is obtained via multiple complex sources or sensors, therefore it would not be wrong to
say that the data that is generated by the IoT system is very raw in the typical terms (Hasan, et.
al., 2019). One of the major characteristics that are associated with IoT is the fact that how
specific the data in IoT has to be but this characteristic only creates the issue of semi supervised
learning as the data that is obtained in the technologies such as IoT or even machine learning are
often unique and considered to be one of their kinds and therefore the chances of having the
assurance by the system that the source of data that is available at the moment is completely

ready for the specification and that makes the developers to label the whole data manually by
their own means. This increases the man power or the cost of the project if instead of manual
labeling, high quality crowd sourcing is used (Hasan, et. al., 2019).
Crowd Sense: It is very well known that crowd sensing is one of the most important aspects in
the process of IoT when it comes to the implementation of Machine Learning in IoT (Predky,
2018). The basic procedure of sensing the crowd is done in 2 ways (Hasan, et. al., 2019). One is
known as Voluntary, where the user of the IoT system voluntarily provide the information to the
system, the other way of crowd sensing is opportunistic, this is the type of crowd sensing in
which the data is obtained automatically and the user has no intervention in that process (Predky,
2018). The main issue that comes in this whole process is that the data that is gathered in the
crowd sensing also consists of a large chunk or a huge amount of unwanted and useless data, and
the task of identifying and removing of that data is way too complex (Hasan, et. al., 2019).

Conclusion
In the document that has been created, the dos and don’ts of internet of things are discussed. The
topic that has been chosen in for document is machine learning in IOT. In the course of this
document a learner would be having the ability to identify and resolve the problems that occur in
the implementation of AI in IoT. Some of the problems that would be discussed in the given
document are Data Consideration, Semi- Supervised Machine Learning, Crowd Sense, Model
Development, Machine leaning in IoT is human centered (Predky, 2018). Internet of Things is
expected to be among the key increment of the latest big trends that are taking place in the new
technologies. By having the collection along the present as well as the estimated ambiguity and
the persuasiveness of the cabled smart systems it provides a base to the applications in various
areas such as the technologies like smart city, smart home, healthcare, transportations,
organizational automations and most importantly retailing . The technical creations have
allowed the internet of things related technology to decrease the price, increase the calculating
ability of the system, cloud service, interoperability and safety and privacies enhancement. One
of the key characteristics that are responsible for the success of internet of things is the
technology of Machine Learning. Machine Learning is among the features of IoT the each and
every system that has been associated with Internet of Things, must be having. But as there have
been some advancements in the technology of IoT as well as Machine Learning there have been
many problems that are observed in associating these 2 technologies together.

References
 Predki, J., (2018). The challenges and opportunities for machine learning in the IoT.
Retrieved From:
https://www.embedded.com/electronics-blogs/say-what-/4460738/2/The-challenges-and-
opportunities-for-machine-learning-in-the-IoT. [Acessed on 16th August 2019]
 Bera, S., Mishra, S., Roy, S., and Obiat, M., S., (2016). Soft-WSN: Software-Defined
WSN Management System for IoT Applications. Institute of Electrical and Electronics
Engineers. 1(1). 1-8
 Khalil, L., Abid, M., R., Benhaddou, D. and Grendt, M., (2016). Wireless Sensor
Network for Internet of Things. Technical University of Munich. 7(1). 1-7
 Jagtaap, S., (2018). IoT + Machine Learning is Going to Change the World. Retrieved
From: https://towardsdatascience.com/iot-machine-learning-is-going-to-change-the-
world-7c4e0cd7ac32. [Acessed on 16th August 2019]
 Hussain, F., Hussain, R., Hassan, S., A. and Hussain, E., (2019). Machine Learning in
IoT Security: Current Solutions and Future Challenges. Retrieved From:
https://arxiv.org/abs/1904.05735. [Accessed on 16th August 2019].