Global Edge School: Notebook entry 2022-23

Key answer – Gr X- AI Project Cycle - TBE

Answer:

Problem
Scoping

Data
Acquisitio
n
Data
Explorin
g
Modelling

Evaluation

Global Edge School – Ahead Of The Learning Curve 1

 Global Edge School: Notebook entry 2022-23

Ans: Problem scoping is part of the project planning stage. This stage allows for establishing
project goals i.e. what has to be achieved? Identifying different stakeholders, the existing
measures for solving the problem and the ethical concerns related to the project.

Global Edge School – Ahead Of The Learning Curve 2

Global Edge School: Notebook entry 2022-23

Global Edge School – Ahead Of The Learning Curve 3

Global Edge School: Notebook entry 2022-23

Global Edge School – Ahead Of The Learning Curve 4

 Global Edge School: Notebook entry 2022-23

1. How can Project evaluation help in boosting the research area?

Ans: Project evaluation helps in finding out the answers to three important questions. First, what
went right, second what went wrong and third what was challenging. The answers to these
questions help in boosting the different research areas. What went right has to be duplicated,
what went wrong has to be corrected and what was challenging has to be simplified. This will in
turn boost the research.

2. How can Project evaluation help in gathering knowledge for accumulation?

Ans: Project evaluation deals with collection of information and its evaluation. This helps
in identifying the correct knowledge for accumulation in different forms. For example,
the successes can be retained for replication and failures can be accumulated for avoidance.

Name some companies that use neural networks and mention how they use it
1. Facebook: The neural network is used for facial recognition.
2. Instagram: The neural network and deep learning is used for identifying the
contextual context of the emoji
3. Amazon: The neural network is used for bringing recommendations to the customers 4.
Google Translate: The neural network is used for automating translations

Global Edge School – Ahead Of The Learning Curve 5

Global Edge School: Notebook entry 2022-23

AI Quiz
QUIZ 3: Activity
Identify the Sustainable Development Goals associated with each of the following
themes:
Environment: SDG 3 (Good Health and Well-Being), SDG 6 (Clean Water and Sanitation),
SDG 7 (Affordable and Clean Energy), SDG 11 (Sustainable Cities and Communities), SDG
12 (Responsible Consumption and Production), SDG 13 (Climate Action), SDG 14 (Life
Below Water), SDG 15 (Life on Land)

Economy: SDG 1 (No Poverty), SDG 2 (Zero Hunger), SDG 7 (Affordable and Clean
Energy), SDG 8 (Decent Work and Economic Growth), SDG 9 (Industrial Innovation
and Infrastructure), SDG 12 (Responsible Consumption and production),

Equality: SDG 1 (No Poverty), SDG 4 (Quality Education), SDG 5 (Gender Equality),
SDG 10 (Reduced Inequalities)
Quiz 4: Identify and Write
You are designing an AI-based smart security system for your school. Identify the
stakeholders for this project.
• Senior Management
• Teachers
• Students
• Parents of students
• Support Workers

Global Edge School – Ahead Of The Learning Curve 6

 Global Edge School: Notebook entry 2022-23

Key answer - AI Project Cycle – NBE

Q1. How is an AI project cycle different from a normal IT project cycle?

Normally, the IT project cycle ends with the deployment phase. Any change after the deployment
phase is treated as a separate project. The AI Projects on the other hand are cyclic in nature i.e.
they do not end with the deployment phase. After deploying the system learns, then testing and
deployment is again repeated.

Q2. Can we do goal setting before scoping the problem? Justify.

Goal setting is one of the stages of problem scoping. The goals of the project are established after
the identification of the problem. These goals set the boundary of the project i.e. what is
and what is not going to be done.

Q3. Can we design AI systems without Data Visualisation? Why?

Strictly speaking data visualisation is not required for designing AI systems. Data
visualisation assist in understanding of data and in recognizing patterns in it. AI cannot
understand data. It can identify patterns, and human beings have to make sense of these
patterns. The use of the data visualisation will depend on the type of data system which is being
designed.

Q4. Are neural networks the best rule-based systems? Why?

Artificial Intelligence systems can either be rules-based systems or they can be learning
based systems. In rules-based systems the selections or processing of data is based on user
defined rules. In neural networks, the selections or processing is based on the learning done by
these networks.

Q5. How is reinforced learning different from supervised learning?

In supervised learning, the AI systems are provided with the problems and the correct solution.
The AI system will establish patterns in the system and use them for finding solutions after
training. In reinforced learning, AI systems do not receive the correct solutions. Rather the right
action or decision is provided with positive reinforcement (positive weight) and incorrect action
or decision is rewarded with negative reinforcement (negative weight).

Q6. Write a short note on learning-based systems.

Global Edge School – Ahead Of The Learning Curve 7

 Global Edge School: Notebook entry 2022-23

Learning based AI systems are not dependent on user established rules for processing data.
These systems rather learn by themselves. They are also known as adaptive intelligence. This
learning can take be supervised, unsupervised or reinforced learning.

Q7. How does the complexity of data influence data visualisation?

Data Visualisation is dependent on the number of variables or parameters involved. Less
complex data is easier to visualise. Complex data will have to be simplified for making
meaningful data visualisation. This can be done through different techniques, data segregation
and classification.

Q8. Write a short note on Big Data.

Big data is the term used for referring to huge amount of data which cannot be stored or
managed in traditional database management systems. Big data is continuously generated in real
life. It can be both structured/unstructured and semi-structured. It can come from various
sources. This data is very important because machine learning systems depend on it.

Q9. How are sustainable development goals linked with artificial intelligence?
Artificial Intelligence can provide significant help in achieving the Sustainable Development
Goals of the United Nations. AI systems among other things can assist in, managing disasters
(SDG 9), creating and maintaining sustainable habitats (SDG 11), assisting the differently abled
(SDG 10), improving productivity (SDG 9), providing better access to education (SDG 4) etc.

Q10. Differentiate between Data Acquisition and Data Exploring.

Data Acquisition deals with collecting or obtaining data from different sources. This is based
on the need on the project. For example, traffic data for traffic lights management systems.
Data Exploration deals on the other hand deals with finding patterns/trends from the
acquired data. The AI system will use these findings for solving the given problems.

Q11. Big data is normally defined on the basis of three parameters. Which are these three
parameters?
The three parameters usually used for defining Big Data are:
Volume: Big Data is a high volume and low-density data. The volume of this data can range from
terabytes to petabytes.

Global Edge School – Ahead Of The Learning Curve 8

 Global Edge School: Notebook entry 2022-23

Velocity: Big data has a high velocity, i.e., this data is received and generated at a faster pace
than traditional data. Many times, Big Data is collected in real-time.
Variety: Big data is made up of structured, unstructured, and semi-structured data. The
proportion of unstructured and semi-structured data usually is much higher than that of
structured data within it. This data needs to be processed before it can be used.

Q12. Why is big data normally unstructured in nature?

Unstructured data is generally defined as data that is not predefined or structured. This data
comes in all shapes, sizes, and forms. It can contain images, audio, videos, and text. Big data
usually is unstructured data because most of the data in the world is unstructured. For example,
data on social media, in newspapers, in emails, etc.

Q13. What is the best source of data for AI System Data Acquisition?
There is no single best source for acquiring data for Artificial intelligence Systems. The selection
of the source depends on the data requirements for the AI system in question. For example, an
AI system designed for helping with navigation will need location and spatiotemporal data for
data acquisition. However, this best data source will be completely useless for an AI system
designed for recognising false news.

Q14. Describe the relationship between data complexity and data visualisation.
There are two relationships that exist between data complexity and data visualisation. On the
one hand, data visualisation becomes difficult with the increase in complexity of the data, and it
becomes comfortable with the decrease in the complexity of data. On the other hand, the
importance of doing data visualisation increases with the increase in complexity of data and
decreases with the decrease in complexity of data.

Q15. What are decision trees? How are they created?

Decision trees are simple models used for classifying data. It is a rule-based AI model which helps
the machine in predicting what an element is with the help of various decisions (or rules) fed to
it. The structure of these models looks like a tree. The decision trees are created with the help of
nodes. The decision tree starts with a root node, which is the first node of the decision tree. It
will involve a test or a question.

Global Edge School – Ahead Of The Learning Curve 9

 Global Edge School: Notebook entry 2022-23

For example, is it raining? Is it Monday? Etc. Each possible answer to the question will result in
one more node. At each node, the AI system uses a test to send the query down a node. This
process continues until the query comes down
to the terminal or leaf node. Each of these terminal and leaf nodes has a preassigned value. This
value is regarded as the output of the decision tree.

Global Edge School – Ahead Of The Learning Curve 10