Artificial Intelligence Assignment April’2020(CSEG344)

Q1. Elaborate the applications of the Artificial Intelligence in the undermentioned domains .
(CO 5)(35 marks )

a. Language Models
Various mathematical and probabilistic methods are accustomed evaluate the likelihood of a
specified series of words appearing during a statement. Word models evaluate text language
bodies and establish the inspiration for his or her term predictions. These are employed
in linguistic communication processing (NLP) applications, specifically those who produce text
as output. a number of these applications include AI and answering questions.
Language models provides a probability that assent could be a legal string during
a language. they're useful as a component of the many NLP systems, like ASR, OCR, and MT.
NLP may be divided into three broad categories. Determined by disruptive ability, these types
are
(i) speech recognition
(ii) linguistic communication understanding, and
(iii) linguistic communication generation.
USES OF MODEL
It is employed in the following-
 Speech recognition
 OCR & Handwriting recognition
 Machine translation
 Context sensitive spelling correction

b. Information retrieval

Data processing deals with the gathering and representation of details and therefore

the processing of knowledge associated with a selected user issue. The knowledge source is
formulating a matter that aims to de-scribe the small print it wants. The specification is
contrasted to the representations within the paper. The descriptions of documents and queries
are normally balanced by a similarity measure, like the Cosine or the Dice coefficient. the
foremost relevant papers are addressed to consumers and are willing to see the importance of
their issues.
Some applications are:
• Web Crawlers (for information extraction)
• Mediator Techniques (for information integration)
• Ontologies (for intelligent information access by making semantics of
• information explicit and machine readable)
Neural Networks (for document clustering & preprocessing)
• Kohonen Neural Networks - Self Organizing maps
• Hopefield Networks
• Semantic Networks

c. Information Extraction

1. Information extraction (IE) is an automatic collection of relevant knowledge linked to a

selected topic from a document object or entity.

2. Data retrieval devices allow it possible to retrieve information from text records,
directories, blogs or various outlets. IE may extract information from unstructured, semi-
structured or structured, electronic text. However, IE is employed in linguistic
communication processing (NLP) to derive organized text from unstructured data. Ex:
Details on oil and gas production.

3. Eg – it may be accustomed extract hate speech from various classification models like

SVM.

d. Natural Language processing –

1. linguistic communication Processing, generally simplified as NLP, could be a subset

of computing that deals with communications between machines and humans using linguistic
communication.
2. the general purpose of the NLP is to interpret, decode, grasp and be of human
languages during a manner that's useful.
3. Many of the NLP strategies specialise in machine learning to infer meaning from human
languages.
4. Example- customer review
• As the foremost crucial aspect, it allows businesses to search out valuable knowledge for his
or her company. It also aims to spice up consumer loyalty.
• If further solutions come up, the foremost appropriate programs are stronger. This also
helps to think about the desires of the patron.
5. Example - Digital digital assistant technologies are probably the foremost common style
of computing.

e. Machine Translation

Machine Translation (MT) is an automatic translation process. this is often the method by which

pc software is employed to translate text from one linguistic communication (e.g. English) to a
different (e.g. Spanish).
In order to perform any conversion, whether human or automatic, the context of the
text within the original (source) language must be entirely restored to the target language, i.e.
the conversion.
Applications of AI
• Text-to-text
• Text-to-speech
• Speech-to-text
• Speech-to-speech
• Image (of words)-to-text

f. Speech Recognition
 Speech recognition could be a tool that may understand spoken sentences, which
may be translated to text. Voice recognition could be a subset of speech recognition, a
system accustomed recognize a individual supported their speech.
 AI made life easier by giving fewer chances of error. It can help to spot the voice of the
tracking of criminals.
 Yahoo, Amazon, Microsoft, Google and Apple — five of the world's biggest software
companies already deliver this functionality on a variety of platforms via services.
The two focus areas we are going to note during this piece are:
o Smart Speaker and Smart Home: Highlighting Amazon, Google and Microsoft
o Mobile Device Applications: Highlighting Apple’s Siri and Facebook’s speech
recognition integrations.
 Again, journalists use the Recordly monitoring method to transcribe speech details to
the published document
 It may be helpful to identify songs dependent on speech inputs.
g. Robot

1. Robots implemented in the commercial field will allow businesses to achieve better and less
mistakes. It is especially beneficial in dynamic production industries such as aerospace. AI
may even be used to help a robot learn on its own which directions are better adapted to
other processes when in service.
2. Robots are now being used for customer service in retail stores and hotels around the
world. Most of these robots use AI's natural language processing capability to communicate
more humanly with customers.
3. Robots also aid with the disinfection of coronavirus sites and are also used to supply
coronavirus patients with medication.
4. Robots can be seen all over the houses, helping with laundry, informing us of our routines,
and even entertaining the babies. The most well-known representation of home robotics is
Roomba's autonomous vacuum cleaner. In fact, robotics have now grown to do everything
from automated mowing lawn to cleaning pools.
5. A.I. Can Speed Up Drug Discovery and Development
6. Drones are being used to monitor situation amid coronavirus
7. A.I. Will be able to forecast epidemics by allowing us adequate time to compile data from
various sources and to track 1000s of infectious diseases.

h. Hardware

Because there are already more forms of AI hardware than ever before, like modern
accelerators, players can deliver easy interfaces and sophisticated AI software application
features, technology needs should change to processing, power, storage, and networking —
and that should turn into various market trends.
Head node-Hardware device that orchestrates and manages computation between accelerators
Accelerator-Silicon chip built to conduct highly parallel operations provided by AI; also allows
simultaneous computation.

 There are a lot of companies operating on Google's AI-specific hardware tensor

processing units (TPUs) that they sell over the cloud and cost only a quarter relative to
training a related platform on AWS.
  Microsoft is investing in Intel's field programmable gate arrays (FGPA) to practice and
infer AI models. FGPA's are extremely configurable, and they can be quickly
programmed and customized.
 Intel has a bunch of hardware for different AI algorithms like CNN. They have purchased
Nervana, a company focused on AI chips, with a strong tech platform for developers as
well.
 IBM is conducting a lot of work on analog computation and phase-change processing for
AI.
 Nvidia dominated the machine learning hardware space because of their fantastic GPUs,
and now they're making things much cooler for AI applications, for example with their
Tesla V100 GPUs.
 There's already a lot of companies like Graphcore, Mythic, and Wave Computing, who
have hundreds of millions of dollars in Venture funding to make preparation and
inferencing easier and cheaper.

i. Perception

Awareness is the method of obtaining, processing, choosing and organizing sensory

input .
There are 2 types of perception -
1) hearing Perception – speech recognition is the front end of a device that can sense
and comprehend spoken language, e.g. speech to speech conversion.
2)vision perception-Real-world perception requires the sensory and cognitive
interpretation of visual information, which in some essential way is close to that usually
experienced during the normal course of everyday human interaction.eg- Object
recognition, object manipulation and navigation.

Eye movement has a major effect on human visual experience. An active vision device is
able to communicate with its surroundings by changing its point of view rather than
actively viewing it, and by working on a series of images rather than on a single
picture.eg – in Mettl through AI based proctoring it can detect phone and other person
detected or you looking away.

j. Planning
 1. Ai does help in making project planning and financial planning better.
2. The planning of Artificial Intelligence involves the decision-making activities undertaken
by machines or computer systems to accomplish a particular purpose.
3. The implementation of the preparation is about selecting a series of acts with a strong
likelihood of achieving a given mission.
4. AI is helpful in financial planning, because the recommendation of the financial advisor
would rely to a large degree on how well the financial planner will keep up with
economic conditions and market data. And on the other side, AI can quickly ingest all
financial historical statistics, assess recent patterns, and link each and every latest
source of knowledge in order to provide an up-to-the-second understanding of markets.
5. AI in project management help in reducing error on each stage ,providing predictive
analysis and solution and improving productivity and efficiency .

k. Moving
AI's ability to track and examine moving objects plays a very important role. Just imagine: smart
drones, cars, robots, sports analytics, contact TV, marketing, advertising — the list of usage
cases is almost endless. The further complex AI artifacts we can control and examine, the more
possibilities we can find. That's why object tracking is so critical.
Watching several items via video is a crucial problem in computer vision. It is used in a number
of video processing applications, such as visual monitoring, sports interpretation, robotic
control, automated driving, human-computer interaction and medical simulation. In the case of
detecting items in a certain type, such as persons or vehicles, the detectors used to make
tracking simpler. Usually, it is done in two steps: Detecting and Tracking.
Object identification is used to conduct a range of AI functions, such as face recognition, vehicle
tracking, health screening, and self-driving. A variety of algorithms depend on target
recognition to operate effectively in AI implementations. The most well known algorithms
linked to object detection processes include R-CNN, Fast RCNN, and Faster RCNN.