Machine Learning for IoT Systems

Machine Learning for Intelligent Decision Making in IoT
Name of the Student
Name of the University
Author’s Note:
Total Words: 6015

Machine Learning for Intelligent Decision Making in
IoT

ABSTRACT – Machine learning refers to the significant field of computer science, which utilizes the statistical
techniques for providing various computer systems, the specific ability in learning with datum, without even getting
explicitly programmed. It is one of the most significant technologies that are utilized in today’s world. It was first
evolved from the various theories of artificial intelligence, computational learning theory and pattern recognition. It
eventually explores the proper construction of algorithms for the purpose of learning to make several vital predictions
about data. Decisions could be easily undertaken with the help of this machine learning. Internet of Things or IoT plays
the most important role in intelligent decision making with machine learning. The main objective of this research report
is to understand the entire concept of machine learning for intelligent decision making in Internet of Things. Various
issues are present within the technology and could be mitigated subsequently. Moreover, the research report also focuses
on the several advantages and disadvantages of the technology.
1. INTRODUCTION (15 MARKS)
The Internet of Things or IoT is defined as the network of
vehicles, home appliances, physical devices or any other
product, which is subsequently embedded with the software,
sensors, electronics, actuators and connectivity, which help in
enabling each and every object in the proper connection and
exchanging of data (Gubbi et al., 2013). Each and every item
can be easily as well as uniquely identified by the specific
embedded system for computing; however comprises of the
core capability to interoperate within the infrastructure of
Internet, which is previously existing. In simple words,
internet of things is the significant interconnection through the
Internet of various computing devices that are embedded
within regular items and thus enabling these objects for
successful sending as well as receiving of data or information
(Kelly, Suryadevara & Mukhopadhyay, 2013). The most
important advantages of this internet of things is that it
increases the machine to machine or M2M connectivity.
Since, it is of ingenious innovation, the physical devices
eventually stay in touch with each and other, hence leading
higher quality and great efficiency.
1.1. MACHINE LEARNING
1.1.1. Background
Machine learning is the specific type of AI or artificial
intelligence, which eventually allows the software
applications in becoming explicitly accurate for the prediction
of outcomes without being perfectly programmed (Lee & Lee,
2015). The main objective of this machine learning is to
construct the algorithms, which could eventually receive the
input data as well as utilize the statistical analysis for
predicting the output value within a proper acceptable range.
Each and every machine learning algorithm is categorized for
getting supervised or unsupervised (He, Yan & Da Xu, 2014).
All of these supervised algorithms need human beings in
providing the input as well as the desired output with the
furnishing feedback regarding prediction accuracy during the
training period. When this training is completed, the specific
algorithm is applied to the new data. The unsupervised
algorithms on the other hand, do not require significant
training for bringing out the desired resulting data. Rather,
they can utilize a specific iterative approach, known as deep
learning for proper data review and thus arriving at
conclusions (Tukker et al., 2013). These unsupervised
algorithms for learning are utilized for tasks that are more
complicated and complex processing than the learning
systems that are supervised.
1.1.2. Machine Learning for Intelligent Decision Making
Machine learning could be easily utilized by all the users
for the purpose of intelligent decision making (Sanchez et al.,
2014). It is the specific cognitive procedure that eventually
results in the perfect selection of any belief or action course
amongst the various alternative possibilities. Each and every
process of decision making subsequently provides the final
choice that might or might not have a prompt action. The
procedure for identification and selection of various
alternatives on the basis of values, beliefs and preferences of
the particular decision maker is called decision making.
Machine learning is one of the most significant technologies
that help to take intelligent decision making (Xu et al., 2014).
The major advantage of involving machine learning in
decision making is that it helps in improving the casual
interference of big data. All these technologies or methods
could be used with any type of complexity and thus are
extremely popular for the end users.
1.1.3. Standout of Machine Learning in Internet of Things
The growth of Internet of Things or IoT has eventually
acquired the entire technological world (Bellavista et al.,
2013). The major involvement of Internet of Things with
machine learning is for the decision making algorithms or
rather for taking various decisions. There are three most
common scenarios, where machine learning is working with
the Internet of Things for enabling the business operations.
The first scenario is anomaly monitoring, where anomalies are
easily detected in shorter time. The second scenario is
predictive maintenance, where the organizational costs are
directly impacted. Thus, the machine learning solution is well
accepted by all (Amendola et al., 2014). The final scenario is
in vehicle telemetry, where the safety as well as reliability of
vehicles is eventually improved. Hence, machine learning is
much popular for the users to take proper decisions.
2. BACKGROUND/LITERATURE REVIEW (40 MARKS)
2.1. Intelligent Decision Making
2.1.1. Data Analytics

According to Guo et al., 2013, IDS or intelligent decision
making system is the specific software package that is utilized
for various important decision analyses. This particular
decision system has the capability in handling several hybrid
kinds of uncertainty that includes interval data, probability
uncertainty, subjective judgments, combination or
amalgamation of all the uncertainty types and even interval
data (Datta, Bonnet & Nikaein, 2014). This intelligent
decision making system utilizes the belief function for the
problem modelling as well as the Evidential reasoning
approach for the attribute aggregation. All the outcomes of
this analysis majorly involve not only the alternative courses
ranking, but also the aggregated performance segregation or
distribution on every alternative to properly support the
transparent and informed decision making (Chi et al., 2014).
The values, preferences as well as the beliefs of the particular
decision maker are selected for the successful procedure of
decision making.
Hassanalieragh et al., 2015 state that data Analytics is the
subsequent procedure for the examination of various data sets
for the purpose of drawing conclusions regarding the
important information, which they contain with the
combination of software and specialized systems (Madakam,
Ramaswamy & Tripathi, 2015). The techniques and
technologies of data analytics that are broadly utilized in any
commercial industry for the sole purpose of enabling the
organizations in making more informative business decisions
and by any researcher or scientist in the proper and significant
verification or disproving of scientific theories, hypotheses
and models. The data analytics eventually refers to the
assortment of the several applications from the basic BI or
business intelligence, OLAP or online analytical processing
and reporting to the several formations of advanced analytics
(Chiang & Zhang, 2016). It is absolutely similar to the nature
of business analytics, which is again one of the major and
significant approaches to analyze data, with the major
difference that this business analytics is properly oriented to
the utilizations of business. This data analytics has the broader
focus. The initiatives of this data analytics could subsequently
helps out any business in increasing the revenues, improving
the operational efficiencies, optimizing marketing campaigns
or any type of customer service effort (Centenaro et al., 2016).
Various competitive advantages are gained or obtained over
the rivals with the help of data analytics.
Figure 1: Entire Process of Data Analytics
The above figure provides a clear image of the process of
data analytics. There are four important stages in the entire
process. The first process starts with the data gathering as well
as processing. This is done with both structured data as well as
unstructured data (Catarinucci et al., 2015). The second stage
is the development of model. In this step, the new patterns are
developed into various models. Moreover, the models are
ranked in this particular step. The third important step in the
process of data analytics is the model testing. All the ranked
models are tested or examined in this process. The final step
of the process is implementation.
As per Baccelli et al., 2014, the intelligent decision making
is done with the help of data analytics and machine learning.
This machine learning plays one of the most significant roles
in the process of decision making. The performance of the
systems and the decision are eventually improvised with the
help of machine learning.
Figure 2: Machine Learning for Decision Making
The above figure describes about the decision making
process with the help of machine learning. There are three
machines in the figure; they are Machine I, Machine II and
Machine III. The three requirements of the decision making
process are large power, low cost and time saving (Fan et al.,
2014). As per the experience or methods, machine I is the
most efficient, machine II is the fastest and machine III
comprises of the greatest power. Finally, the application is
applied and the results are drawn.

2.1.2 Risk Intelligent Decision Making
According to Jiang et al., 2014, risk intelligent decision
making refers to that specific type of decision making that
help to manage the risks of the business and that decision.
Any organization or business or individual always has the
major risk to take any type of decisions within their business
(Desai, Sheth & Anantharam, 2015). These risks could be
extremely dangerous and threatening for the business and thus
there is a high chance the business or organization can
undergo various problems for the business. There are few
skills in this risk intelligent decisions making that are to be
considered (Ishaq et al., 2013). The major skills mainly
include maintaining vigilance, by which the special detecting
mechanisms are set up and thus developed a proper range of
the potential responses and managing vital connections or
links that reduces the complexity and interconnectedness of
the entire global business environment for making it difficult
and tougher for looking how each event could affect each
other (Thirumalai & Kar, 2017). The next significant skill that
is required for managing the risks in the risk intelligent
decision making is anticipation of the causes of failure. The
potential failure could be recognized easily and quickly for
escalating the remediation to a specific level. The corroborate
information and the sources should be verified properly and
the safety and security should be maintained eventually
(Zhang et al., 2014). Moreover, few risks are to be taken for
understanding the situation properly.
As per Biswas and Giaffreda, 2014, Risk Intelligent
decision making is much easier with the help of machine
learning. The amount as well as chance of risks is explicitly
reduced with this machine learning. The most important and
significant advantages of this machine learning for reducing
the risks or managing the risks and taking significant
decisions mainly involve fast processing as well as real
predictions, the churn analysis, customer defections and many
more. Various industries could easily implement machine
learning for reducing their business risks properly and hence it
is extremely popular for all users (Yannuzzi et al., 2014). It is
proactive in nature and provides data accuracy, which refers to
the fact that the decisions that are taken with this machine
learning are absolutely accurate and perfect for the business.
Figure 3: Process of Decision Making with Machine
Learning
The above figure has clearly shown the entire process of
decision making with the help of machine learning. Since,
decision making is an important phenomenon for any
business; it should be properly utilized in all the organizations
(Pang, 2013). In the figure, the first step is giving input. The
data with no dependency on the perception of artificial
intelligence or AI results is checked in the first step. With the
help of the perception or the first step, the decision is being
taken in the machine learning engine (Farooq et al., 2015).
The most popular types of artificial intelligence are speech
recognition, image recognition, voice analytics and video
analytics. This decision making layer of artificial intelligence
in the second step then provides the final step of perfect
output. Due to this perfect output, the decision is also perfect
and accurate and hence all types of risks are easily avoided in
the process (Zhang et al., 2013).
2.1.3. Decision Making in IoT
Xu, Wendt and Potkonjak, 2014, state that The supply
chains are extremely complex and complicated systems,
which require short as well as precise concepts for the people
for taking significant decisions. The perfect implementation of
any innovation is extremely challenging and thus the
influencing factors as well as the impacts on these supply
chains are broad ranging (Cirani et al., 2015). In today’s
world, the supply chains are eventually dependent on the
collection of data by those equipments that are Internet
connected for improving the efficiency of the operations. IoT
or the Internet of Things subsequently contributes to the
capability in gathering all these data and finally combining the
processes, people and equipments (Krco, Pokric & Carrez,
2014). The use of Internet of Things is much more than any
other technology in the modern world. The various
applications and the wear able are much popular for the
Internet of Things. The process of decision making comprises
of various threats and vulnerabilities and thus they have
several positive as well as negative impacts. Hence, there is a
well defined approach in making accurate and appropriate
decisions (Li et al., 2013). The decision making process is
eventually enhanced by the technology of Internet of Things.
The ability in analyzing the perfection of decisions and how
this Internet of Things influences the decisions of business are
noted in a proper manner. Firstly, making decisions is part of
every moment in life and thus a topic of which everyone
should have at least basic knowledge available. Secondly,
Internet of Things is an emerging topic with a promising
future effecting personal and business life equally (Yashiro et
al., 2013). Finally, supply chain management enables
competitive advantages if correctly managed and executed.
The research methodology of this research helps in defining
or finding the recent model as well as strategies of the field
that are being taken from various sources within the fields of
businesses and science (Margelis et al., 2015). There are
various approaches that prove the fact the decisions taken by
the machine learning are either qualitative or quantitative.
Both of these approaches are extremely important for the
business or decision maker and the emphasis is given on the
subjective findings as well as the individual findings. This
type of approach thus leads to various answers. On the other
hand, quantitative approaches in decision making are much
easier and simpler in comparison to the qualitative approach
(TongKe, 2013). From a cognitive more fundamentally point
of view, researchers assume the nature of decisions as a

merger of different possibilities and impact rates while the
decision maker, as executive of decisions, being able to
influence the impact by having ready several strategies and
approaches (Amadeo et al., 2014). In order to find the
appropriate decision among the vastness of information,
alternatives (at least two needs to be given), separation and
differentiation need to be conducted by these strategies.
According to Pa et al., 2014, there are various sample
technologies of Internet of Things that help the user in taking
various decisions for the users. The solutions of the SAP
Internet of Things solely ensure or simplify the user’s
interactions with the devices. The best example of this internet
of things is Intel IoT platform (Liu et al., 2015). This
particular platform is responsible for enabling the exchange,
storage as well as analysis in the complete flow of data. The
first or the initiation point is a sensor and the ending point is
the data center with specific focus on the security of
transmission as well as the compatibility within or across the
businesses. The platform of Intel IoT eventually helps the
customers in improving the device integration, system
organizations and data organizations (Gaur et al., 2015). The
customer relationship management or CRM and the enterprise
resource planning or ERP systems are properly integrated with
the help of this Internet of Things. The mobility features of
any organization are eventually facilitated for the core purpose
of making the networks properly visible as well as monitored
in such a way that the decisions could be easily undertaken by
the organization (Theodoridis, Mylonas & Chatzigiannakis,
2013). The communication of devices and the mutual
recognition are much more easier with the help of this Internet
of Things or IoT.
Figure 4: Internet of Things
The above figure shows the various products of Internet of
Things like phone, computers and many more. All of these are
extremely important and significant to understand the entire
concept of Internet of Things or IoT for any decision maker.
2.2 Machine Learning
2.2.1. Authentication and Trust Issues
As per Riahi et al., 2-13, the technology of machine
learning is one of the most important and significant part in
today’s world. Machine learning is thus considered as the
dominant approach in the field of AI or artificial intelligence.
In recent days, machine learning is dominating the fields of
computer vision, speech recognition, computer dialogue
systems, robotic controls and natural language question
answering (Karimi & Atkinson, 2013). Moreover, computer
science areas such as detection of spam, consistency of
databases and retrieval of information are also done with the
help of machine learning. The entire procedure of construction
of computer software is easily done with the help of this
machine learning. Machine learning has the significant
relationship in the field of neurosciences (Truong & Dustdar,
2015). As per various researches, the current understanding of
this neuroscience is extremely poor in respect to the value of
learning mechanisms. The artificial neural networks are
directly linked to the actual neurons and are the most
important tools within the learning mechanism engineering.
There is a basic difference the specialized learning and the
generalized learning (Amadeo, Campolo & Molinaro, 2014).
The specialized learning refers to the fact that there are
various specialized systems that help in the learning
phenomenon and the general learning refers to the fact that
there are various technologies that help in education of the
students.
However, in spite of having all these advantages in various
sectors, there are some of the major disadvantages (Blaauw et
al., 2014). There are various authentication issues as well as
trust issues for this type of artificial intelligence parts. The
first and the foremost issue that machine learning comprises
of and is dangerous for any decision making in Internet of
Things is requirement of larger working memory. The
memory that is utilized in the machine learning technology
should be extremely larger and then only data could be stored
(Al-Fuqaha et al., 2015). If the memory size is not large, the
data will not be stored and thus the decision maker cannot
take up the decision without the help of adequate data. The
neural network is required to be properly hooked up to the
specific memory block, which could be both read well as
written by the typical network. This is one of the major
authentication issues in the machine learning (Santoso & Vun,
2015). The next significant trust issue within the technology
of machine learning is NLP or natural language processing
failure. There is always a high chance that this particular type
of language processing can be a massive failure even for the
deeper networks. Since, the languages are represented and the
reasoning is simulated, this becomes a major authentication
issue and the decision maker faces significant problem in the
phenomenon. The third important trust issue of the machine
learning is training of the deep networks (Gubbi et al., 2013.
The real progress could not be measured in this case and thus
the deep networks are required to be trained properly. This
particular problem of machine learning is considered as one of

the most important issue that is required to be mitigated as
soon as possible.
According to Kelly, Suryadevara and Mukhopadhyay,
2013, apart from the above mentioned trust issues in machine
learning, there are some other issues as well, which disturb the
authentication of this particular technology. One shot learning
is one of them. For the successful achievement of one shot
learning, there are certain applications of the neural networks
that are to be evolved (Lee & Lee, 2015). However, this one
shot learning has not been properly achieved by the users. Due
to these issues, the decision maker suffers through major
issues while taking such decisions. The larger amount of data
is required to be extracted and thus this to be done properly
and significantly. Another important authentication issue is
the deep reinforcement technique for learning in controlling
the robots. The semantic segmentation is yet another trust
issue in machine learning and thus difficult for the decision
makers in taking the decisions (He, Yan & Da Xu, 2014). The
detection of objects is also difficult for the users and the thus
they suffer from various problems for the users.
Figure 5: Difference between Machine Learning and Deep
Learning
The above figure demonstrates a clear image between the
machine learning and deep learning. Machine learning is
much more advanced and proper for the users and thus should
be properly utilized by the decision makers for perfect
decision making in internet of things (Tukker et al., 2013). A
significant feature extraction is present in machine learning.
However, this feature learning is absent in deep learning.
3. ISSUES/ SOLUTIONS (10 MARKS)
3.1. Automation Issue in Machine Learning
Machine learning being one of the most important and
significant technology in modern world comprises of various
issues or challenges within the technology (Sanchez et al.,
2014). The first and the foremost issue that is present within
this technology is the issue of automation. The processes that
require automation are the most important in the entire
scenario of machine learning. Automation can be defined as
the significant technology, through which any specific process
or procedure is being performed without any type of human
assistance (Bellavista et al., 2013). This automation or simply
automatic control is the utilization of several control systems
to the operating equipments like machinery, procedures in the
factories, and ovens for heat treating or transferring, switching
on the telephone networks, boilers, switching or changing on
the telephone networks, stabilization as well as steering of
ships or aircrafts.
This automation is responsible for covering the applications
that are ranging from the household thermostat maintaining or
controlling any boiler to the larger system of industrial control
with numerous input measurements as well as signals of
output control. Within the complexity of the output control, it
can eventually range from the simplified on off control to the
multi variable algorithms that are higher level (Datta, Bonnet
& Nikaein, 2014). This particular issue of automation
becomes a major threat for the decision making in the
technology of internet of things. The simple or the easiest
processes or procedures that are to be automated is required to
be evaluated the type of problems, the decision makers are
requiring to solve.
3.1.1. Solution to Automation Issue
However, this particular problem of process automation
could be easily resolved if the processes are executed properly
(Biswas & Giaffreda, 2014). The platform of artificial
intelligence is to be selected or utilized properly for mitigating
this specific type of issue in the decision making process. The
problems are to be evaluated properly for resolving the issues.
The most easiest or the simplest processes are required or
needed to be automated. These processes carry manual data or
manual output. The complex or the complicated processes or
procedures are required for further introspection before the
automation is even done. Machine learning can definitely help
automate some processes, not all automation problems need
machine learning (Pang, 2013). This particular solution to the
problem of automation in machine learning would eventually
enhance the area of the decision making and the process of
decision making extremely easier for the decision makers
within the technology of Internet of Things.
3.2. Inadequate Infrastructure
The second important problem or issue that is common for
the machine learning for decision making in Internet of
Things technology is inadequate infrastructure. Machine
learning needs huge or vast amount of the capability of data
churning. Hence, the legacy systems cannot handle the
workload and thus buckle under pressure (Desai, Sheth &
Anantharam, 2015). As Internet of Things or IoT are involved
in the scenario, the requirement of infrastructure is huge and
thus the user often faces major problems or issues. Moreover,
if this infrastructure is not upgraded or does not get updated in
every means, the data or information is eventually lost and
there is no scope of recovery.
3.2.1 Solution to Inadequate Infrastructure
For mitigating this particular problem of inadequate
infrastructure, the decision maker should check for the
infrastructure regularly and thus the machine learning could
be controlled or handled properly. This machine learning can
only work properly if the systems or infrastructures are
working perfectly. For this issue, the decision maker should
implement upgraded systems so that they can get accurate
data or information (Yannuzzi et al., 2014). Moreover, regular
up gradation of the systems and flexibility in storage would
help the decision maker eventually. The acceleration of

hardware is yet another important requirement of machine
learning and this would reduce the complexity of inadequate
infrastructure and thus helping the user of Internet of Things
in taking proper and accurate or intelligent decisions.
3.3. Data Problems
The most significant issue for the machine learning
technology or any decision maker is the problem of data (Chi
et al., 2014). Data is the most important and vital resource that
is required by any user or decision maker to take proper
decisions in any specific domain. If the data or information
will not be properly present for the user or decision maker,
they will not be able to take the accurate decision for the
business. These data problems are thus extremely vital and
require to be resolved as soon as possible. The quality of data
is required to be the best of all and for that various algorithms
are to be implemented. This particular issue of data problems
can be for various types of data like incomplete data, dirty
data and noisy data (Chiang & Zhang, 2016). These are the
quintessential enemies of ideal machine learning.
3.3.1. Solution to Data Issue
This particular issue of the data problem could be easily
resolved with the help of proper measures. The solution to this
specific problem is taking the time of the decision maker for
evaluation as well as scope of data through the meticulous
governance of data, integration of data as well as the
exploration of data (Datta, Bonnet & Nikaein, 2014). The data
then becomes perfect and proper and thus the decision could
be easily taken with the help of machine learning. The major
issue of data quality could be resolved if the input of the data
is perfect and for this the decision maker should undergo
various qualitative and quantitative approaches.
4. FUTURE RESEARCH (5 MARKS)
4.1. Gap Analysis and Various Future Directions
The technology of machine learning has eventually
acquired the entire technological world with the various
advantages or benefits. From detection of fraud cases to the
self driving cars, this particular part of the artificial
intelligence is considered as the biggest technological boon
for its users (Xu, Wendt & Potkonjak, 2014). There is a lot of
scope of this particular technology in the future and the future
directions of this technology are explicitly fascinating for the
users throughout the world.
4.1.1. Quantum Computing
The first future direction of the machine learning
technology is the quantum computing. Machine learning tasks
involve problems such as manipulating and classifying large
numbers of vectors in high-dimensional spaces. Quantum
computers would likely be extremely good at manipulating
high-dimensional vectors in large tensor product spaces (Krco,
Pokric & Carrez, 2014). It is likely that both the development
of both supervised and unsupervised quantum machine
learning algorithms will hugely increase the number of vectors
and their dimensions exponentially more quickly than
classical algorithms. This would eventually result in the huge
increase within the speed at which all the algorithms of
machine learning will run.
4.1.2. Collaborative Learning
The second future direction for the machine learning in
Internet of things is the collaborative learning. This is the
significant procedure regarding the utilization of various
computational entities for the reason they could collaborate
for producing better learning result than these had been
achieved on their own (Margelis et al., 2015). The most
significant example of this collaborative learning is the
usability of the nodes of the Internet of things or IoT sensor
networks. The other name of this network is edge analytics.
With the explicit growth of the IoT, it is likely that large
numbers of separate entities will be utilized to learn
collaboratively in many ways.
Figure 6: Collaborative Learning
The above figure depicts a clear image of the advantages of
collaborative learning. The six parts of this collaborative
learning mainly involve sharing ideas, interaction,
brainstorming, community sharing, discussion and finally
collaboration of the various parts.
4.1.3. Cognitive Services
The third important future direction of the machine learning
technology is the cognitive services. This particular
technology mainly involves the kits such as APIs or
application programming interfaces as well as services
through which any developer could create more intelligent and
discoverable applications (TongKe, 2013). The machine
learning application programming interfaces or APIs would
eventually enable the developers in introducing the various
intelligent features like the detection of speech, face,
emotions, visions, voice and many more. The future of this
field will be the introduction of deeply personalized
computing experiences for all.
Figure 7: Cognitive Services of Machine Learning

The above figure depicts a clear image of the cognitive
services of the future of machine learning. The entire process
of decision making is much easier with this technology and
thus this is yet another important future direction of machine
learning (Amadeo et al., 2014). Thus, the decision makers can
easily take up various decisions without any type of
complexity, even if the decision is related to Internet of
Things.
5. ADVANTAGES/ DISADVANTAGES (5 MARKS)
5.1. Advantages of Machine Learning in IoT
Internet of Things is the proper network of the various
physical devices that are subsequently embedded with the
electronics, sensors, connectivity, software and many more
that helps to enable all the objects in proper connection as
well as exchanging of data or information. Each of the things
is properly recognized within the specific embedded
computing system; however has the core capability in
interoperating within the significant existing infrastructure of
Internet (Liu et al., 2015). When this Internet of Things is
linked with the machine learning, there are some of the major
advantages or benefits that help the user to take various
significant decisions. The most significant advantage that the
user gets from machine learning in IoT is that the user gets
maximum data and better predictive analytics with the proper
machine learning.
The deep learning requires some of the major conversion
for the decision makers (Theodoridis, Mylonas &
Chatzigiannakis, 2013). The big data is leveraged and thus
Internet of Things or IoT is optimized properly. Machine
learning is regarding the search of the useful patterns for the
provided data set. These usually identify correlations between
input values that you can observe, and output values that you'd
eventually like to predict. The first and the foremost
advantage of machine learning for any decision maker in
internet of things is the data input from unlimited or
unrestricted resources. There is no limit of the resources and
thus the decision maker can utilize this particular feature
within their machine learning. The second advantage of
machine learning is that it does not incur any type of
complexity and thus could be easily utilized by all users in
practical scenarios (Karimi & Atkinson, 2013). Due to the
lack of complexity, machine learning is utilized within the
medical industry. Another important advantage of this
machine learning for the intelligent decision making in
internet of things is that it is extremely proactive and it
designed for the action as well as reaction industries. All the
systems could easily act within the machine learning output.
5.2. Disadvantages of Machine Learning in IoT
The technology of machine learning do comprise of several
disadvantages or demerits for intelligent decision making in
internet of things (Karimi & Atkinson, 2013). The first
disadvantage is that is comprises of acquisition. There are
various data algorithms that are to be processed. It thus has a
proper impact of the results that are to be achieved or
obtained. Interpretation is the next important disadvantage of
machine learning for the decision makers. The effectiveness
of the algorithms of machine learning is checked properly.
The third limitation of the machine learning in Internet of
Things is the limitation. This makes the machine learning
technology a major failure. Another significant disadvantage
in machine learning for decision making in IoT is the error
susceptibility (Truong & Dustdar, 2015). The correction and
the diagnosis of the machine learning is extremely difficult
and thus this is required to be mitigated with proper mitigation
measures. The final disadvantage of this particular technology
for the decision makers within the internet of things is that
there is a lack of variability.
6. CONCLUSION (5 MARKS)
Therefore, from the research report on Machine Learning
for Intelligent Decision Making in Internet of Things, it can be
concluded that machine learning is the significant application
of AI or artificial intelligence, which eventually provides
various systems, the core capability of automatic learning as
well as improving from previous experience. This does not
involve any type of programming within it. The machine
learning majorly focuses on the proper growth or development
of various computer programs, which could easily access or
utilize the data for the purpose of learning. The procedure of
learning starts with data or observations like examples of
direct experience and instructions for looking for the patterns
within data and thus making the decisions much better in all
the future based examples. The main objective of the machine
learning is the enabling of several computers in learning
automatically. The most vital advantage of the machine
learning is that it does not require any human intervention or
assistance in the process and thus adjusting all the actions as
per order. Machine learning is explicitly involved in the IoT
or Internet of Things. The main feature of this combination is
that they help in taking significant decisions for the users.
Some of the major methods of machine learning are
supervised machine learning algorithms, unsupervised
machine learning algorithms, semi supervised machine
learning algorithms and reinforcement machine learning
algorithms. These are sub divided into either supervised or
unsupervised machine learning algorithms. For the purpose of
decision making, the users can easily combine or mix the two
extreme and significant technologies of Internet of Things and
machine learning. There are five stages when an important
decision is being taken with the help of machine learning and
Internet of Things. The five stages are event production, event
ingestion, transformation and analytics, persistence or storage
and presentation or actions. The main point of this specific
process and what drives the significant business value is being
encapsulated within the third stage of the specific activity
chain. This third stage is transformation and analytics. In this
particular stage, the data is inspected and thus proper
decisions are taken. All these decisions directly influence the
activities, which would be optimizing the flow of businesses.
The capability of the systems in making cognitive decisions
that are on the basis of historical data; majorly influence the
solution value and importance. Various technologies such as
Azure Machine learning could easily leverage the techniques

for supervised learning for helping the businesses in making
several business decisions or individual decisions on the basis
of regression, detection of anomalies and classification. The
primary cause of the re emergence of this machine learning is
the proper evolution of the machine computing and the
adoption of the technological world. The most significant
advantages of machine learning for intelligent decision
making within Internet of Things are high scalability, high
storage and computing capabilities, higher performing
services of computing, flexibility, pay as per usability model
of subscription and many more. In spite of having such
significant advantages, there are some of the major
disadvantages of this machine learning that make the user
suffer in taking various decisions. The future of the machine
learning in Internet of Things is extremely high and has the
capability in acquiring the technological world properly. The
above research report has properly outlined all the important
points and details of the technologies of machine learning and
internet of things for the purpose of taking significant
decisions and helping the users and the technological world.
Moreover, the various advantages, disadvantages and the
future analysis of these technologies and the amalgamation of
these technologies are eventually and properly provided here.
References:
Al-Fuqaha, A., Khreishah, A., Guizani, M., Rayes, A., &
Mohammadi, M. (2015). Toward better horizontal integration
among IoT services. IEEE Communications Magazine, 53(9),
72-79.
Amadeo, M., Campolo, C., & Molinaro, A. (2014, September).
Multi-source data retrieval in IoT via named data networking.
In Proceedings of the 1st ACM Conference on Information-
Centric Networking (pp. 67-76). ACM.
Amadeo, M., Campolo, C., Iera, A., & Molinaro, A. (2014, June).
Named data networking for IoT: An architectural perspective.
In Networks and Communications (EuCNC), 2014 European
Conference on (pp. 1-5). IEEE.
Amendola, S., Lodato, R., Manzari, S., Occhiuzzi, C., & Marrocco,
G. (2014). RFID technology for IoT-based personal healthcare
in smart spaces. IEEE Internet of things journal, 1(2), 144-152.
Baccelli, E., Mehlis, C., Hahm, O., Schmidt, T. C., & Wählisch, M.
(2014, September). Information centric networking in the IoT:
experiments with NDN in the wild. In Proceedings of the 1st
ACM Conference on Information-Centric Networking (pp. 77-
86). ACM.
Bellavista, P., Cardone, G., Corradi, A., & Foschini, L. (2013).
Convergence of MANET and WSN in IoT urban
scenarios. IEEE Sensors Journal, 13(10), 3558-3567.
Biswas, A. R., & Giaffreda, R. (2014, March). IoT and cloud
convergence: Opportunities and challenges. In Internet of
Things (WF-IoT), 2014 IEEE World Forum on (pp. 375-376).
IEEE.
Blaauw, D., Sylvester, D., Dutta, P., Lee, Y., Lee, I., Bang, S., ... &
Yoon, D. (2014, June). IoT design space challenges: Circuits
and systems. In VLSI Technology (VLSI-Technology): Digest of
Technical Papers, 2014 Symposium on (pp. 1-2). IEEE.
Catarinucci, L., De Donno, D., Mainetti, L., Palano, L., Patrono, L.,
Stefanizzi, M. L., & Tarricone, L. (2015). An IoT-aware
architecture for smart healthcare systems. IEEE Internet of
Things Journal, 2(6), 515-526.
Centenaro, M., Vangelista, L., Zanella, A., & Zorzi, M. (2016).
Long-range communications in unlicensed bands: The rising
stars in the IoT and smart city scenarios. IEEE Wireless
Communications, 23(5), 60-67.
Chi, Q., Yan, H., Zhang, C., Pang, Z., & Da Xu, L. (2014). A
reconfigurable smart sensor interface for industrial WSN in IoT
environment. IEEE transactions on industrial
informatics, 10(2), 1417-1425.
Chiang, M., & Zhang, T. (2016). Fog and IoT: An overview of
research opportunities. IEEE Internet of Things Journal, 3(6),
854-864.
Cirani, S., Picone, M., Gonizzi, P., Veltri, L., & Ferrari, G. (2015).
Iot-oas: An oauth-based authorization service architecture for
secure services in iot scenarios. IEEE sensors journal, 15(2),
1224-1234.
Datta, S. K., Bonnet, C., & Nikaein, N. (2014, March). An IoT
gateway centric architecture to provide novel M2M services.
In Internet of Things (WF-IoT), 2014 IEEE World Forum
on (pp. 514-519). IEEE.
Desai, P., Sheth, A., & Anantharam, P. (2015, June). Semantic
gateway as a service architecture for iot interoperability.
In Mobile Services (MS), 2015 IEEE International Conference
on(pp. 313-319). IEEE.
Fan, Y. J., Yin, Y. H., Da Xu, L., Zeng, Y., & Wu, F. (2014). IoT-
based smart rehabilitation system. IEEE transactions on
industrial informatics, 10(2), 1568-1577.
Farooq, M. U., Waseem, M., Khairi, A., & Mazhar, S. (2015). A
critical analysis on the security concerns of internet of things
(IoT). International Journal of Computer Applications, 111(7).
Gaur, A., Scotney, B., Parr, G., & McClean, S. (2015). Smart city
architecture and its applications based on IoT. Procedia
computer science, 52, 1089-1094.
Gubbi, J., Buyya, R., Marusic, S., & Palaniswami, M. (2013).
Internet of Things (IoT): A vision, architectural elements, and
future directions. Future generation computer systems, 29(7),
1645-1660.
Guo, B., Zhang, D., Wang, Z., Yu, Z., & Zhou, X. (2013).
Opportunistic IoT: Exploring the harmonious interaction
between human and the internet of things. Journal of Network
and Computer Applications, 36(6), 1531-1539.
Hassanalieragh, M., Page, A., Soyata, T., Sharma, G., Aktas, M.,
Mateos, G., ... & Andreescu, S. (2015, June). Health monitoring
and management using Internet-of-Things (IoT) sensing with
cloud-based processing: Opportunities and challenges.
In Services Computing (SCC), 2015 IEEE International
Conference on (pp. 285-292). IEEE.
He, W., Yan, G., & Da Xu, L. (2014). Developing vehicular data
cloud services in the IoT environment. IEEE Transactions on
Industrial Informatics, 10(2), 1587-1595.
Ishaq, I., Carels, D., Teklemariam, G. K., Hoebeke, J., Abeele, F. V.
D., Poorter, E. D., ... & Demeester, P. (2013). IETF
standardization in the field of the internet of things (IoT): a
survey. Journal of Sensor and Actuator Networks, 2(2), 235-
287.
Jiang, L., Da Xu, L., Cai, H., Jiang, Z., Bu, F., & Xu, B. (2014). An
IoT-oriented data storage framework in cloud computing
platform. IEEE Transactions on Industrial Informatics, 10(2),
1443-1451.
Karimi, K., & Atkinson, G. (2013). What the Internet of Things (IoT)
needs to become a reality. White Paper, FreeScale and ARM, 1-
16.
Kelly, S. D. T., Suryadevara, N. K., & Mukhopadhyay, S. C. (2013).
Towards the implementation of IoT for environmental condition
monitoring in homes. IEEE Sensors Journal, 13(10), 3846-
3853.
Krco, S., Pokric, B., & Carrez, F. (2014, March). Designing IoT
architecture (s): A European perspective. In Internet of Things
(WF-IoT), 2014 IEEE World Forum on (pp. 79-84). IEEE.

Lee, I., & Lee, K. (2015). The Internet of Things (IoT): Applications,
investments, and challenges for enterprises. Business
Horizons, 58(4), 431-440.
Li, F., Vögler, M., Claeßens, M., & Dustdar, S. (2013, June).
Efficient and scalable IoT service delivery on cloud. In Cloud
Computing (CLOUD), 2013 IEEE Sixth International
Conference on (pp. 740-747). IEEE.
Liu, C., Yang, C., Zhang, X., & Chen, J. (2015). External integrity
verification for outsourced big data in cloud and IoT: A big
picture. Future Generation Computer Systems, 49, 58-67.
Madakam, S., Ramaswamy, R., & Tripathi, S. (2015). Internet of
Things (IoT): A literature review. Journal of Computer and
Communications, 3(05), 164.
Margelis, G., Piechocki, R., Kaleshi, D., & Thomas, P. (2015,
December). Low throughput networks for the IoT: Lessons
learned from industrial implementations. In Internet of Things
(WF-IoT), 2015 IEEE 2nd World Forum on (pp. 181-186).
IEEE.
Pa, Y. M. P., Suzuki, S., Yoshioka, K., Matsumoto, T., Kasama, T.,
& Rossow, C. (2015). IoTPOT: analysing the rise of IoT
compromises. EMU, 9, 1.
Pang, Z. (2013). Technologies and Architectures of the Internet-of-
Things (IoT) for Health and Well-being (Doctoral dissertation,
KTH Royal Institute of Technology).
Riahi, A., Challal, Y., Natalizio, E., Chtourou, Z., & Bouabdallah, A.
(2013, May). A systemic approach for IoT security.
In Distributed Computing in Sensor Systems (DCOSS), 2013
IEEE International Conference on (pp. 351-355). IEEE.
Sanchez, L., Muñoz, L., Galache, J. A., Sotres, P., Santana, J. R.,
Gutierrez, V., ... & Pfisterer, D. (2014). SmartSantander: IoT
experimentation over a smart city testbed. Computer
Networks, 61, 217-238.
Santoso, F. K., & Vun, N. C. (2015, June). Securing IoT for smart
home system. In Consumer Electronics (ISCE), 2015 IEEE
International Symposium on (pp. 1-2). IEEE.
Theodoridis, E., Mylonas, G., & Chatzigiannakis, I. (2013, July).
Developing an iot smart city framework. In Information,
intelligence, systems and applications (iisa), 2013 fourth
international conference on (pp. 1-6). IEEE.
Thirumalai, C., & Kar, H. (2017, April). Memory Efficient Multi
Key (MEMK) generation scheme for secure transportation of
sensitive data over Cloud and IoT devices. In Power and
Advanced Computing Technologies (i-PACT), 2017 Innovations
in (pp. 1-6). IEEE.
TongKe, F. (2013). Smart agriculture based on cloud computing and
IOT. Journal of Convergence Information Technology, 8(2).
Truong, H. L., & Dustdar, S. (2015). Principles for engineering IoT
cloud systems. IEEE Cloud Computing, 2(2), 68-76.
Tukker, A., de Koning, A., Wood, R., Hawkins, T., Lutter, S.,
Acosta, J., ... & Kuenen, J. (2013). EXIOPOL–development and
illustrative analyses of a detailed global MR EE
SUT/IOT. Economic Systems Research, 25(1), 50-70.
Xu, B., Da Xu, L., Cai, H., Xie, C., Hu, J., & Bu, F. (2014).
Ubiquitous data accessing method in IoT-based information
system for emergency medical services. IEEE Transactions on
Industrial Informatics, 10(2), 1578-1586.
Xu, T., Wendt, J. B., & Potkonjak, M. (2014, November). Security of
IoT systems: Design challenges and opportunities.
In Proceedings of the 2014 IEEE/ACM International
Conference on Computer-Aided Design (pp. 417-423). IEEE
Press.
Yannuzzi, M., Milito, R., Serral-Gracià, R., Montero, D., &
Nemirovsky, M. (2014, December). Key ingredients in an IoT
recipe: Fog Computing, Cloud computing, and more Fog
Computing. In Computer Aided Modeling and Design of
Communication Links and Networks (CAMAD), 2014 IEEE 19th
International Workshop on (pp. 325-329). IEEE.
Yashiro, T., Kobayashi, S., Koshizuka, N., & Sakamura, K. (2013,
August). An Internet of Things (IoT) architecture for embedded
appliances. In Humanitarian Technology Conference (R10-
HTC), 2013 IEEE Region 10 (pp. 314-319). IEEE.
Zhang, Y., Raychadhuri, D., Ravindran, R., & Wang, G. (2013). ICN
based Architecture for IoT. IRTF contribution, October.
Zhang, Z. K., Cho, M. C. Y., Wang, C. W., Hsu, C. W., Chen, C. K.,
& Shieh, S. (2014, November). IoT security: ongoing challenges
and research opportunities. In Service-Oriented Computing and
Applications (SOCA), 2014 IEEE 7th International Conference
on (pp. 230-234). IEEE.