© 2018 IJNRD | Volume 3, Issue 9 September 2018 | ISSN: 2456-4184

MACHINE LEARNING (ML)

The trending topic in information technology field
[1]
Afroz Labbai, [2]Arbaaz Iqbal Nakhwa , [3]Khan Ibrahim Firoz, [4]Uzair Khalil Faquih
[1]
Student at ai-arkp, [2]Student at ai-arkp, [3]Student at ai-arkp, [4]Student at ai-arkp.
[1]
Computer Department of Ai-arkp.
[1]
Anjuman-I-Islam's Abdul Razzaq Kalsekar Polytechnic ,Navi-mumbai,India.

I. Abstract: This paper is an information article on Machine  Reinforcement machine learning algorithms is a learning
Learning. Machine learning is an trending topic of IT method that interacts with its environment by producing
industry. It is developed in mostly python, java, R, c++, actions and discovers errors or rewards
javascript, scala language. Ml is used in a wide range in our
day to day life. And Ml is going to change many things in the II. How Ml and AI changed The industry
future. Mostly Ml is used for BigData problem solution, Artificial Intelligence and Machine Learning have been forecasted to
Pattern learning, etc. Ml is purely algorithm based working. be the game-changers of the coming decade. Most of the tech trends
discussed really fall under the umbrella field of Data Science, as they
II. Keywords: AI, IoT, VR, AR, Big Data, Chatbot, Analytics, directly or indirectly affect Data Management. However, some
Algorithm, ML. specific innovations like Augmented Reality[9] (AR), Virtual
Reality[10] (VR), or Digital Twins also indicate hardware
enhancements and overlapping of the physical and digital worlds.
III. Introduction to ML(Machine Learning): In a blog post tiled Forrester’s Top Emerging Technologies To
Machine learning is a field of computer science that uses statistical Watch: 2017-2021, Forrester made strong assertions about an
techniques to give computer systems the ability to "learn" (e.g., impending disruptive technology world, where suddenly an
progressively improve performance on a specific task) with data, avalanche of ground-breaking digital technologies such as mobile,
without being explicitly programmed. social, Big Data, Cloud, IoT[8], Ai, and ML will all emerge and
The name machine learning was coined in 1959 by Arthur converge to turn the global businesses upside down.
Samuel.[5] Evolved from the study of pattern In another Forrester’s post tiled The Top Technology Trends To
recognition and computational learning theory in artificial Watch: 2016 To 2018, the author claims that Big Data and IoT[8] have
intelligence[6] , machine learning explores the study and construction suddenly catapulted the need for Intelligent Analytics systems
of algorithms that can learn from and make predictions on data – powered by robotic assistants and self-teaching algorithms. What
such algorithms overcome following strictly static program this trend indicates is that the future business owners, leaders, and
instructions by making data-driven predictions or decisions, through managers will have more time to invest on critical business issues as
building a model from sample inputs. Machine learning is employed day-to-day Data Management tasks will be taken over by intelligent
in a range of computing tasks where designing and programming machines. The additional layer of ML-driven AI capabilities has
explicit algorithms with good performance is difficult or infeasible; taken the afore-mentioned technologies to the ultimate era of
example applications include email filtering, detection of network technological freedom, where man and machine co-habitate happily.
intruders, and computer vision. The transformative outcomes of such a “connected business world”
Machine learning is an application of artificial intelligence[6] (AI) are participative customer relationships, data-powered decision
that provides systems the ability to automatically learn and improve making, implementation of innovative risk and security strategies,
from experience without being explicitly programmed. Machine and superfast technology maturity driven by global customer
learning focuses on the development of computer programs that can obsession with everything tech. The businesses, big or small should
access data and use it learn for themselves. take advantage of this era of digital disruptions and empower their
The process of learning begins with observations or data, such as workforces and customers alike.
examples, direct experience, or instruction, in order to look for In the next five years, say by 2021, the fully automated,
patterns in data and make better decisions in the future based on the intelligent Data Management systems may actually end up chopping
examples that we provide. The primary aim is to allow the computers about 6 percent of US-based Data Science jobs. In 2021, we may
learn automatically without human intervention or assistance and even see a virtually connected, business ecosystem once again
adjust actions accordingly transforming the customer engagement platforms and methods.

I. Some machine learning methods: III. What Major Industry Watchers Think About
 Supervised machine learning algorithms can apply what Artificial Intelligence and Machine Learning
has been learned in the past to new data using labelled According to the column Gartner, IDC, and Forrester on the Future
examples to predict future events of Digital Transformation:
 Semi-supervised machine learning algorithms fall “By 2018, 35% of IT resources will be spent to support the creation
somewhere in between supervised and unsupervised of new digital revenue streams and by 2020 almost 50% of IT
learning, since they use both labelled and unlabelled data budgets will be tied to digital transformation initiatives. Enterprises
for training – typically a small amount of labelled data and a pursuing digital transformation initiatives will more than double the
large amount of unlabelled data.

IJNRD1809007 International Journal of Novel Research and Development (www.ijnrd.org) 35

 © 2018 IJNRD | Volume 3, Issue 9 September 2018 | ISSN: 2456-4184

size of their software development teams by 2018, focusing new 3. Videos Surveillance:
hires almost entirely on digital initiatives.” Imagine a single person monitoring multiple video cameras!
The CXO Today article 7 AI Technologies To Power The Certainly, a difficult job to do and boring as well. This is why the
Enterprise indicates that Ericsson has candidly claimed in its idea of training computers to do this job makes sense. The video
6th Annual Trend Report that about 35 percent of advanced internet surveillance system nowadays are powered by AI that makes it
users prefer an AI advisor and about 25 percent hope to see an AI possible to detect crime before they happen. They track unusual
Manager next year. behaviour of people like standing motionless for a long time,
In Top Technology and IT Predictions for 2017 and Beyond, the IT stumbling, or napping on benches etc. The system can thus give an
industry leader HCL proclaims that industry watchdogs like IDC, alert to human attendants, which can ultimately help to avoid
Gartner, or Forrester have all claimed that the 2017 will usher in an mishaps. And when such activities are reported and counted to be
era of “accelerated digital disruption” making 2018 more true, they help to improve the surveillance services. This happens
technologically significant than ever. The top technologies that will with machine learning doing its job at the backend.
make this era happen are AI[6], ML, and IoT[8]. In the newly
connected business world, intelligent systems may increasingly 4. Social Media Services:
collaborate, engage, and solve business problems over disconnected People You May Know: Machine learning works on a simple concept
geographies. This is where the virtually connected world will understanding with experiences. Facebook continuously notices the
triumph over physical boundaries. friends that you connect with, the profiles that you visit very often,
your interests, workplace, or a group that you share with someone
Some of the common forecasts made by the above industry watchers etc.
for 2018 and the coming years are: Face Recognition: You upload a picture of you with a friend and
1. The IT trend of newer digital services with additional Facebook instantly recognizes that friend. Facebook checks the
revenue streams has started and will gain momentum in poses and projections in the picture, notice the unique features, and
2018 before reaching 40 percent in 2019. then match them with the people in your friend list.
2. The tech trend of “connected” and intelligent systems has
started and will gain momentum in 2018 before we see 40 5. Online Customer Support:
percent of digital transformation and upwards of 70-80 A number of websites nowadays offer the option to chat with
percent of AI-powered, IoT[8] initiatives in 2019. customer support representative while they are navigating within the
3. By 2022, IoT[8] will save businesses and customers an site. However, not every website has a live executive to answer your
astronomical sum of $1 trillion a year in maintenance and queries. In most of the cases, you talk to a chatbot[7]. These bots tend
services. Some of that will start happening in 2018. to extract information from the website and present it to the
4. In 2018, more analytics vendors will offer IoT [8]-driven customers. Meanwhile, the chatbots[7] advances with time. They tend
capabilities which may lead global businesses to invest to understand the user queries better and serve them with better
more in AI[6]-powered, connected technologies. According answers, which is possible due to its machine learning algorithms.
to Forrester, automated IoT [8] has already been adopted in Machine learning is an application of artificial intelligence (AI [6])
transportation industry fleet management, government that provides systems the ability to automatically learn and improve
security and surveillance systems, retail inventory from experience without being explicitly programmed. Machine
management, and manufacturing assets management. learning focuses on the development of computer programs that can
access data and use it learn for themselves.
IV. Application of Machine Learning: The process of learning begins with observations or data, such as
1. Virtual Personal Assistants: examples, direct experience, or instruction, in order to look for
Siri, Alexa, Google Now are some of the popular examples of virtual patterns in data and make better decisions in the future based on the
personal assistants. As the name suggests, they assist in finding examples that we provide. The primary aim is to allow the computers
information, when asked over voice. learn automatically without human intervention or assistance and
adjust actions accordingly.
Virtual Assistants are integrated to a variety of platforms. For
example: V. Some machine learning methods:
 Smart Speakers: Amazon Echo and Google Home Machine learning algorithms are often categorized as supervised or
 Smartphones: Samsung Bixby on Samsung S8 unsupervised. Supervised machine learning algorithms can apply
 Mobile Apps: Google Allo what has been learned in the past to new data using labeled examples
to predict future events. Starting from the analysis of a known
2. Predictions while Commuting: training dataset, the learning algorithm produces an inferred function
Traffic Predictions: We all have been using GPS navigation services. to make predictions about the output values. The system is able to
While we do that, our current locations and velocities are being saved provide targets for any new input after sufficient training. The
at a central server for managing traffic. This data is then used to build learning algorithm can also compare its output with the correct,
a map of current traffic. While this helps in preventing the traffic and intended output and find errors in order to modify the model
does congestion analysis, the underlying problem is that there are accordingly.
less number of cars that are equipped with GPS. Machine learning in In contrast, unsupervised machine learning algorithms are used when
such scenarios helps to estimate the regions where congestion can be the information used to train is neither classified nor labeled.
found on the basis of daily experiences. Online Transportation Unsupervised learning studies how systems can infer a function to
Networks: When booking a cab, the app estimates the price of the describe a hidden structure from unlabeled data. The system doesn’t
ride. When sharing these services, how do they minimize the figure out the right output, but it explores the data and can draw
detours? The answer is machine learning. inferences from datasets to describe hidden structures from
unlabeled data.
IJNRD1809007 International Journal of Novel Research and Development (www.ijnrd.org) 36
 © 2018 IJNRD | Volume 3, Issue 9 September 2018 | ISSN: 2456-4184

Semi-supervised machine learning algorithms fall somewhere in segment text topics, recommend items and identify data outliers.
between supervised and unsupervised learning, since they use both Semi supervised learning is used for the same applications as
labeled and unlabeled data for training – typically a small amount of supervised learning. But it uses both labeled and unlabeled data for
labeled data and a large amount of unlabeled data. The systems that training – typically a small amount of labeled data with a large
use this method are able to considerably improve learning accuracy. amount of unlabeled data (because unlabeled data is less expensive
Usually, semi-supervised learning is chosen when the acquired and takes less effort to acquire). This type of learning can be used
labeled data requires skilled and relevant resources in order to train it with methods such as classification, regression and prediction. Semi
/ learn from it. Otherwise, acquiring unlabeled data generally doesn’t supervised learning is useful when the cost associated with labeling
require additional resources. Reinforcement machine learning is too high to allow for a fully labeled training process. Early
algorithms is a learning method that interacts with its environment by examples of this include identifying a person's face on a web cam.
producing actions and discovers errors or rewards. Trial and error Reinforcement learning is often used for robotics, gaming and
search and delayed reward are the most relevant characteristics of navigation. With reinforcement learning, the algorithm discovers
reinforcement learning. This method allows machines and software through trial and error which actions yield the greatest rewards. This
agents to automatically determine the ideal behavior within a specific type of learning has three primary components: the agent (the learner
context in order to maximize its performance. Simple reward or decision maker), the environment (everything the agent interacts
feedback is required for the agent to learn which action is best; this is with) and actions (what the agent can do). The objective is for the
known as the reinforcement signal. Machine learning enables agent to choose actions that maximize the expected reward over a
analysis of massive quantities of data. While it generally delivers given amount of time. The agent will reach the goal much faster by
faster, more accurate results in order to identify profitable following a good policy. So the goal in reinforcement learning is to
opportunities or dangerous risks, it may also require additional time learn the best policy.
and resources to train it properly. Combining machine learning with Humans can typically create one or two good models a week;
AI and cognitive technologies can make it even more effective in machine learning can create thousands of models a week.
processing large volumes of information.
Reference:
VI. Statistics and Machine Learning at Scale [1]: Name designation, etc of the 1st author.
Those driverless cars we keep hearing about from Google? They’re [2]: Name designation, etc of the 2nd author.
just one example of machine learning – and we’re about to hear of [3]: Name designation, etc of the 3rd author.
many more. Even though the concept of machine learning has been [4]: Name designation, etc of the 4th author.
around for decades, it’s now becoming more pertinent to our daily [5]: Samuel, Arthur (1959)”Some studies in machine learning using
lives because of increased amounts of data and cheaper storage. the game of checkers”. IBM journal of research and development.
This paper, which is based on a presentation given at the Analytics [6]: Artificial intelligence, sometimes called machine intelligence, is
Conference, introduces key machine learning concepts and intelligence demonstrated by machines, in contrast to the natural
introduces SAS solutions – SAS In-Memory Statistics for Hadoop intelligence displayed by humans and other animals.
and SAS Visual Statistics – that enable machine learning at scale. [7]: A chatbot (also known as a talkbot, chatterbot, Bot, IM bot,
interactive agent, or Artificial Conversational Entity) is a computer
VII. Some popular Machine Learning methods program or an artificial intelligence which conducts a conversation
Two of the most widely adopted machine learning methods are via auditory or textual methods.
supervised learning and unsupervised learning – but there are also [8]: IoT is short for Internet of Things. The Internet of Things refers
other methods of machine learning. Here's an overview of the most to the ever-growing network of physical objects that feature
popular types. an IP address for internet connectivity, and the communication that
Supervised learning algorithms are trained using labeled examples, occurs between these objects and other Internet-enabled devices and
such as an input where the desired output is known. For example, a systems.
piece of equipment could have data points labeled either “F” (failed) [9]: Augmented Reality (AR) is an interactive experience of a
or “R” (runs). The learning algorithm receives a set of inputs along real-world environment whereby the objects that reside in the
with the corresponding correct outputs, and the algorithm learns by real-world are "augmented" by computer-generated perceptual
comparing its actual output with correct outputs to find errors. It then information, sometimes across multiple sensory modalities,
modifies the model accordingly. Through methods like including visual, auditory, haptic, somatosensory, and olfactory.
classification, regression, prediction and gradient boosting, [10]: Virtual reality (VR) is an
supervised learning uses patterns to predict the values of the label on interactive computer-generated experience taking place within a
additional unlabeled data. Supervised learning is commonly used in simulated environment, that incorporates mainly auditory and visual,
applications where historical data predicts likely future events. For but also other types of sensory feedback like haptic.
example, it can anticipate when credit card transactions are likely to
be fraudulent or which insurance customer is likely to file a claim.
Unsupervised learning is used against data that has no historical
labels. The system is not told the "right answer." The algorithm must
figure out what is being shown. The goal is to explore the data and
find some structure within. Unsupervised learning works well on
transactional data. For example, it can identify segments of
customers with similar attributes who can then be treated similarly in
marketing campaigns. Or it can find the main attributes that separate
customer segments from each other. Popular techniques include
self-organizing maps, nearest-neighbor mapping, k-means clustering
and singular value decomposition. These algorithms are also used to
IJNRD1809007 International Journal of Novel Research and Development (www.ijnrd.org) 37