0% found this document useful (0 votes)
41 views6 pages

CAT 1 - Artificial Intelligence Programming

Uploaded by

Gabriel Gitonga
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
41 views6 pages

CAT 1 - Artificial Intelligence Programming

Uploaded by

Gabriel Gitonga
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 6

CAT 1– Artificial Intelligence Programming

GABRIEL GITONGA
21/08375
Question 1:
a) Define machine learning and describe its main categories: Supervised,
Unsupervised, and Reinforcement learning. (3 Marks)
Machine Learning (ML) is a branch of artificial intelligence that systems to learn from
data and gradually get better at what they do without code being written explicitly.
Machine learning can be divided into the following main categories:

Supervised Learning: In this method, the input data is paired with the correct output, and
the model is trained on a labelled dataset. The goal is for the model to learn a mapping
from inputs to outputs.

Unsupervised Learning: In this case, unlabelled data is used by the model. Without
specifying the desired outputs, it finds patterns or structures in the data.

Reinforcement learning: An agent is trained to make decisions by acting in a way that


increases accumulating rewards. Through feedback and trial and error, the participant
gains knowledge based on their actions

b) Provide one real-world example for each category, explaining how ML is applied
in that scenario. (3 Marks)
Email Spam Detection: Supervised Learning: Machine learning algorithms analyse
emails classified as "spam" or "not spam" to identify traits of spam emails. The learnt
patterns can then be used by the trained model to categorize newly received emails.

Unsupervised Learning: client Segmentation: Without established categories, retailers


employ unsupervised learning to examine client purchase records. Businesses can
customize their advertising strategies for various categories of customers by finding
groups of people who have similar purchasing habits.

Reinforcement Learning: Autonomous Vehicles: Reinforcement learning is used by


self-driving cars to navigate and make judgments. By rewarding appropriate driving
practices and penalizing risky ones, the car learns from its surroundings and gradually
improves its driving technique.
c) Explain what Natural Language Processing (NLP) is and outline its importance
in AI applications.
(2 Marks)
Natural Language Processing (NLP) is a field of artificial intelligence that focuses
with how people and computers communicate using natural language. It has to do
with machines' comprehension, interpretation, and meaningful response to human
language.

Importance in AI Applications:

Large volumes of natural language data may be processed and analyzed by machines
thanks to NLP, which enhances human-machine communication.
It enhances user experiences and makes technology more accessible by powering
apps like chatbots, virtual assistants, and translation services.

d) Describe how AI-powered NLP systems like chatbots are transforming customer
service in modern industries. (2 Marks)
Chatbots and other AI-powered natural language processing (NLP) systems are
transforming customer service by offering 24/7, immediate assistance. They can handle
numerous questions and issues without the need for human interaction, understanding and
reacting to consumer requests in real time. As a result, wait times reduce, efficiency is
increased, and human agents can concentrate on more difficult jobs. Chatbots, for instance,
can help with troubleshooting, order tracking, and FAQs, all of which increase customer
happiness.

e) Explain the role of computer vision in AI, and discuss how it is used in
healthcare for diagnosing diseases. Provide an example.
(2 Marks)
Machines can now understand and read visual data from the outside world, including
pictures and movies, thanks to the artificial intelligence field of computer vision.

Computer vision is used in the medical field to assess medical pictures, such as CT, MRI,
and X-rays, in order to help diagnose illnesses. AI can help radiologists diagnose diseases
like tumors or fractures more quickly and correctly by spotting abnormalities and patterns
in these images.

As an example, Google's DeepMind created an artificial intelligence system that examines


eye scans to find early signs of diabetic retinopathy, a condition which may prevent
blindness by allowing immediate treatment.
Question 2: Propositional Logic and Reasoning
a) Define propositional logic and explain its significance in AI, especially in
automated decision-making systems like expert systems.

(3 Marks)
Propositional Logic is a branch of logic that analyses statements that might or might not
be true. It uses logical conjunctions like AND, OR, and NOT to construct complicated
arguments from simpler ones.

Propositional logic is important for artificial intelligence (AI) since it is used by expert
systems and other machines that use standards formed by logical statements to reach
decisions or draw conclusions. It provides an understandable basis for thinking about
knowledge and enables the automation of complicated processes of decision-making.
.

b) Discuss how propositional logic can be used to automate decision-making in either


a medical diagnosis or legal reasoning scenario. (2 Marks)
By creating rules based on symptoms and situations, propositional logic may speed up
decision-making in medical diagnosis. For example, a patient may have disease X if they
have symptoms A and B. An expert system can compare patient data to these rules and
make suggestions for potential diagnoses by employing a sequence of logical statements.

By analysing legal statements and conditions, propositional logic in legal reasoning can
assist in automating case analysis. An AI system may assess events to determine whether
the requirements for legal actions are fulfilled, for example, if a contract specifies that "If
A occurs, then B must follow," which would help lawyers make decisions

c) Compare forward and backward chaining in AI reasoning systems. Provide a real-


world example of how each is used in practical applications. (3 Marks)
Chaining Forward:
In order to come at a conclusion, this technique of reasoning begins with known facts and
uses rules to derive new facts.
As an example, forward chaining in an expert medical diagnosis system starts with the
patient's symptoms and uses rules to determine potential illnesses based on those
symptoms.

Chaining backwards:
This approach determines which truths must be true in order to achieve a goal by starting
with the goal and working backward.
Backward chaining, for instance, may begin with the objective of establishing the
innocence of a person and go backward through the law to ascertain which evidence is
necessary for supporting that conclusion in a legal reasoning framework.
d) Briefly explain why propositional logic alone is not sufficient for handling complex
real- world AI problems, and suggest what logical system might overcome these
limitations.
(2 Marks)

Because it is unable to manage complicated relationships, uncertainty, or the dynamic


character of real-world events, propositional logic has weaknesses. Since it only works
with propositions that are true or untrue, it is difficult to communicate knowledge that
involves details like probability or vagueness.

Predicate logic, sometimes referred to as first-order logic, is a logical framework that can
get around these limitations and support more complicated statements involving objects
and their relationships. Fuzzy logic is also appropriate for real-world applications because
it can manage uncertainty by permitting truth values to vary between true and false.
Question 3: Applications of AI
a) Discuss how AI is improving diagnostic accuracy and treatment recommendations in
healthcare. Include one AI tool or system used in hospitals. (3 Marks)
AI in healthcare helps in the analysis of patient data, such as test results, medical pictures,
and medical history, to help physicians make more precise diagnoses. By accessing huge
medical databases and comparing a patient's condition to similar cases, it can also suggest
individualized treatment approaches.

IBM Watson for Oncology is an example of a medical AI technology. It assists physicians


in making customized decisions about cancer care by recommending cancer therapy
options based on a large body of clinical data and research.

b) Explain how AI-driven recommendation systems work, and describe their


impact on customer experience in e-commerce platforms like Amazon or
Netflix.
(2 Marks)
Retail recommendation systems driven by AI examine consumer information, including
browsing history, previous purchases, and preferences, to propose goods or content that
the user is likely to find interesting. By providing individualized recommendations, this
enhances customer pleasure and engagement on sites like Amazon and Netflix, helping
to customize the viewing or purchase experience. For instance, Amazon makes product
recommendations based on past purchases or things that are often bought together,
while Netflix makes show recommendations based on a user's viewing preferences.

c) Describe the architecture of a simple neural network. Use a labeled diagram to


explain how input data is processed through different layers. (3 Marks)

Neural networks process data through a structure of interconnected nodes (neurons) that mimic the
functionality of the human brain. Each neuron receives input data, applies an activation function,
and forwards the output to the subsequent layer.

The architecture typically consists of:


1. Input Layer: The entry point for input data.
2. Hidden Layers: These layers execute computations and extract features from the
data. The complexity of the problem dictates the number of hidden layers.
3. Output Layer: Generates the final output, which could be a prediction or
classification.
Input Layer Hidden Layer (1) Hidden Layer (2) Output Layer
(Features) (Neurons) (Neurons) (Prediction)
[X1] ──► [H1] [H3] [Y]
[X2] ──► [H2] [H4]
[X3] ──►
[X4] ──►

Processing Steps:

1. Input Data: Raw data enters the input layer.


2. Forward Propagation: The data passes through hidden layers, where weights are updated
based on learned patterns.
3. Activation Function: Each neuron applies an activation function (e.g., ReLU for non-
linear activation) to decide whether to send its signal forward.
4. Output Generation: The output layer produces the final prediction based on the processed
input data.

END

You might also like