YVETTE RUTH

19/06448
BBIT (AI PROJECT)
TOWN (CAMPUS FULL TIME)
QUESTION ONE
Artificial intelligence, as known as AI, is the imitation or cloning of human intelligence, that allows
machines, more specifically computer systems, to perform in an extremely intelligent manner. What
exactly does it do then? So basically, an artificially intelligent system perceives its environment in a
manner that helps its analytical self in maximizing its chances of success. John McCarthy coined the term
in 1956 and was defined as the science and engineering of making intelligent machines. It emphasizes
on developing machines to think and work just like humans. For each of the following scenarios describe
how you would develop an intelligent system to solve problems.
Every day we get dozens of email notifications and most of them are spam. Build a tool which can
classify the emails as spam or non-spam based on the content of the email alone. To achieve this, I
would build a tool that uses spam filtering techniques that I will discuss below
1a.
(I) Adaptive Spam Filtering Technique
Using this, your emails will be detected and grouped into different classes. Each group will
be assigned an emblematic text. As emails come in, a comparison of it will be made to each
group then a percentage of similarity will dertimine the group in which the email belongs to
thus grouped.
(II) Previous likeness based spam filtering technique
This technique uses the k nearest neighbor for filtering spam emails. As the technique
name suggests, we will be using already stored example emails to classify incoming emails
based on their ressemblance and will be using methods such as machine learning, memory
based or instance based
(III) Content based filtering technique
This technique uses machine learning approaches such as neural networks, support vector
machine, k nearest neighbor to create automatic filtering rules to classify emails. This ethod
will analyse phrases and words used in the emials the will filter the incoming spam emais
(IV) Case base spam filtering method
This method uses machine learning algorithm to train data sets and test them inorder to
decide whether the incoming mail is spam or not.first all the emails both spam and non
spam are all collected then undergoes processes such as selection, feature extraction,
grouping of email data and evaluating the process then the data is classified into two vector
sets.
(V) Heurstic or rule based spam filtering technique
This method usually uses regular expressions against a chosen message using already
created rules or heuristics to asses a huge number of patterns. Several similar patterns
increase the score of a message

1b) One of the challenging tasks of AI is to find the optimal path from one place to the
destination place. The idea is to find the optimal path for a vehicle to travel so that cost and
time can be minimized. This is a business problem that needs solutions
The best way to go about this is by the use of RPS ( intelligent route planning system). RPS is
part of a project called INSIGMA ( intelligent information system for global monitoring,
detection and identification of threats) what it does is that it detects, identifies and
monitors objects in an environent such as road networks and public areas.

The system will calculate the best route to use using various types of the optimization
algorithms which will take into consideration availability of traffic information and user
preference.the later is obtained by use of many types of detectors such as microwave
sensors, gps trackers, acoustic arrays, video detectors.

1c) A chatbot system that can negotiate the price of the products. Many eCommerce
companies are researching on this project as it has a lot of scope in business. The chatbot
should be able to communicate with users and handle the bargaining of customers just like
in the real world For this, I would use AI powered bots specifically the three main types
which iw ill discuss below;

(I) Smart bots

However the name, such bots are not perfect and their work is not fully
automated since they need some form of human intervention to answer very
specific customer inquiries, the still provide a better positive user expirience
since they intereact with the users in a human like way. These chatbots are
powered by AI
(II) Script bots
They are the most popular ecommerce chat bots. They function according to
predefined scripts and commands. They are intergrated with NLP ( natural
language processing) to make interactions more natural with the user. They are
programmed to recognize words and answer basic questions
(III) Intelligent agent
This type combines both AI technologies and machine learning which makes
the fully self sufficient, they are the most advanced type of chatbots.

1e) Lane line detection technique is used in many self-driving autonomous

vehicles as well as line-following robots. We can use computer vision
techniques such as color thresholding to detect the lanes We can use heuristic
method clustering methodology to group detected points and best fit line in the
mean squre sense to detect the lanes. Many sensors can be used for lane
detection such as radar and vision sensors and laser sensors. These will be used
to detect the obstacles on the road by the use of inverse perspective mapping
method(IPM) the system will obtain the front view using camera which is
mounted on the vehicle then applies processes inorder to detect lanes and
objects. We can then install in the vehicle intelligent systems such as road lane
detection and also obstacle detection which will make it easier. Principle
approaches to detection are using vision systems on the vehicle.

Discuss the AI concepts that can be used in the following aspects [25 Marks] a.
Artificial Intelligence for a Better World Many people say that technology is
snatching away their jobs and with the machine, there is no need for humans.
But, do you know that it is these machines that are making the world a better
place to live in. It is this AI, that is helping us to prevent future damage. It
understands the needs and addresses developmental needs while focusing on
sustainability. Do you know that companies like Microsoft are using AI to study
land-use patterns with terrain maps? By understanding these patterns in-depth,
better decisions related to land are taken. This is helping in implementing
proper preservation techniques. Scientists are using the information obtained
to preserve biodiversity and the ecosystem. Discuss the various ways in which
AI techniques can be used in land use management to enhance sustainability.
(I) Smart agricultural and food systems
AI augumented agriclture involves automated dat collection
decision manking and corresponding corrective measures via
robotics indoer to allow early detection of crop diseases, proide
time nutrition to plants and generally to optimise agriculral inputs
and returns based on supply and demand. This promises to increase
the resource efficiency of the agriculture industry lowering the use f
water fertilizers and pestisides which causes damge to important
ecosystems

(II) Smart disaster response

AI can analyse real time and simulation data of whether events and
disaster to seek out vunrablities and enhance disaster preparation,
prioritise early response. . Deep reinforcement learning may one
day be integrated into disaster simulations to determine optimal
response strategies, similar to the way AI is currently being used to
identify the best move in games like AlphaGo

(III) Reinforcing learning for earth science breakthroughs

This nascent AI technique – which requires no input data,

substantially less computing power, and in which the evolutionary-
like AI learns from itself – could soon evolve to enable its
application to real-world problems in the natural sciences.
Collaboration with Earth scientists to identify the systems – from
climate science, materials science, biology, and other areas – which
can be codified to apply reinforcement learning for scientific
progress and discovery is vital

(iV)Distributed energy grids

AI can enhance the predictability of demand and supply for
renewables across a distributed grid, improve energy storage,
efficiency and load management, assist in the integration and
reliability of renewables and enable dynamic pricing and trading,
creating market incentives

2b
(I) BAGGING REGRESSION TREE
Bagging comes from bootstrap aggregation. This is a
method of selecting a random number of data from a
set with replacement. It is mostly to reduce the
variance of the tree.

(II) Linear regression To determine the correlation

between two continuous variables, simple linear
regression analysis is used. One of the two variables is
the predictor variable of which value is to be found. It
gives the statistical relationship not the deterministic
relationship between two variables. Linear regression
algorithm gives yhe best fit line to the given data for
which the prediction error is minimum. Gradient
descent and cost function are the two major factors to
understand linear regression
(III)
Decision tree This tree count isolates the collected
information into small subsets, at a comparative same
time makes it persistent. The last results show the tree
having the decision centres, likewise, the leaf centres.
This decision centre point may contain two branches at
any rate. At first think about the whole informational
index as root. Feature regards are kicked out of the
chance. In case the characteristics are relentless then
they have to be discretized before structuring the
model. In view of estimation property records are
corrected recursively

(IV) Support vector machine The SVM is supervised ML

algorithm is used for classification and regression
analysis. It takes a large time to process so usually
applied to the small datasets. It finds the Hyperplane
that separates the feature into different parts. It gives
optimal hyperplane which classifies the different
domains. The data points which are nearest to
hyperplane are called support vector points and
distance between the vector plane and these points are
called as margins. The proposed work has exploited
SVM for regression analysis. The performance depends
on kernel function selection as a nonparametric
technique. Linear, Radial Basis Function and Polynomial
are the kernels of support vector machine algorithm
(V)K nearest neighbors s, the output is mean of its k
nearest neighbours. Like SVM this is also a
nonparametric method. Considering few values, results
are computed to achieve the best value. KNN is a
supervised classification algorithm that can also be
used as a regressor. It assigns a new data point to the
class. It is non-parametric because it does not take any
assumption. It calculates the distance between every
training example and a new data point