Mount Abu Public School

Practical Record
Subject: Artificial Intelligence
Subject Code: 843

Name: Saanvi Raju

Class: XI
Section: A
Session: AISSCE 2023 - 24

Teacher’s Signature
 Table of Contents
Sl. Topic Name Page Date
No. Number

Part Unit 1 - Introduction to AI for

A everyone

1 AI for All – Certificate 1 12.08.2021

2 AI for All – Badge 2 12.08.2021

3 AI for Oceans - 3 23.08.2021

★ Introduction
★ Certificate

4 Supervised Learning and 6 26.08.2021

Reinforcement Learning Activity
through Artbot

5 Neural Network Concept through 9 27.08.2021

Quick Draw

6 AI Domain – Data 11 02.09.2021

7 AI Domain - Computer Vision 13 03.09.2021

8 CV - Shadow Art 17 04.09.2021

9 AI Domain – NLP 19 07.09.2021

10 Teachable Machine Activity 23 31.08.2021

★ Image Project ★
Audio Project
★ Pose Project
PART A - INTRODUCTION TO AI FOR EVERYONE

1. AI for All - Certificates

2. AI for All - Badges

 3. AI for Oceans

★ Introduction

AI for Oceans is a lesson designed by code.org in which we get to classify

objects on the basis of if they are “fish” or “not fish” in order to help the AI
bot to remove trash from oceans. We also teach the bot to include other sea
creatures and apply particular labels to fishes. After teaching the AI bot, it
should be able to differentiate the objects the way we have taught it.

This activity employs the concept of Machine Learning*, especially

Supervised Learning▲ in which training data is used to enable a Machine
Learning Model (here, the AI Ocean Bot) to classify new data. This also
includes an understanding of AI Bias∘.

*Machine Learning – It is a subset of AI which enables a computer

system to learn from experience instead of programming it.
“Machine learning is the science of getting computers to act without being
explicitly programmed.”
- Stanford University
▲
Supervised Learning – It happens in the presence of a supervisor or a
teacher. The machine is trained with labelled data (which is already tagged
with correct answers). Then, the machine is prepared to predict the
outcomes of unforeseen or new data based on the learnt information it
gained during the training process.
∘
AI Bias – It is an issue that arises when an AI algorithm generates results
that are systematically prejudiced due to inaccurate assumptions in the
machine learning process.
AI for Ocean Screenshots:
★ Certificate
 4. Supervised Learning and Reinforcement Learning
Activity through Artbot

Art Bot - It is an interactive web-based game in which the quest is to find

and retrieve stolen art objects. We train the AI helper to recognise and
locate the objects hidden in a maze of dungeons. This employs supervised
learning and reinforcement learning.

Supervised Learning – It happens in the presence of a supervisor or a

teacher. The machine is trained with labelled data (which is already tagged
with correct answers). Then, the machine is prepared to predict the
outcomes of unforeseen or new data based on the learnt information it
gained during the training process.

Reinforcement Learning – It is an area of machine learning in which the

machine is not given examples of correct input-output

ArtBot Screenshots:
 5. Neural Network Concept through Quick Draw

Quick Draw is a game built with machine learning which uses Neural
Network Concept* to recognise doodlings or drawings made by the user.
It learns with experience and gets better when users add their drawings to
the world’s largest doodling data set. This is shared publicly to help with
machine learning research.

*Neural Network Concept - Neural Network, also known as Artificial

Neural Network or ANN, is a series of algorithms that can identify
underlying relationships in a set of data by following the way the human
brain operates. This is based on the concept of the system of
(natural/biological) neurons in humans, and hence the concept of artificial
neurons is born and put into effect.
Quick Draw Screenshots:
 6. AI Domain - Data

Data: A data domain is the collection of values such as raw facts, sets of
numbers and characters, images, videos, etc. AI systems can handle huge
volumes of data and instead of processing, they recognise data and its
patterns which helps them in predicting the result with similar kinds of
data.

Computer games that are based on the AI domain of Data are Tic-Tac-Toe
and Rock Paper Scissors.

★ Tic Tac Toe: It is a retro-version of the simple game of X’s and O’s
which we can play with the computer. It is based on the AI domain of
Data. It recognises the pattern in our moves and eventually starts defeating
us almost every time.

★ Rock Paper Scissors: It is a classic rock-paper-scissors game against

AI. The computer is able to predict our moves based on our history
because we are very likely to repeat ourselves. It identifies patterns within
the data.
The more we play, the more data is collected and the AI gets better at
predicting our moves.

➔
2D Rock-Paper-Scissors

➔
3D Rock-Paper-Scissors
 7. AI Domain - Computer Vision

Computer Vision (CV): It is a field of artificial intelligence (AI) of

machines that helps to analyse visual data (like digital images, videos and
others), derive meaningful information and patterns from them and
eventually, make decisions, take actions or give recommendations based on
that information received. Computer vision enables machines to see,
observe and understand the world.

Computer games that are based on the AI domain of Computer Vision are
Teachable Snake and Emoji Scavenger Hunt.

★ Teachable Snake: Teachable Snake is an interactive web game.

Instead of using physical buttons to control the game, a black arrow drawn
on paper can be used as a controller to move the snake by turning the paper
in different directions in front of the webcam.
★ Emoji Scavenger Hunt: This online game challenges us to locate the
emoji it shows in the real world with our phone’s camera. A neural
network will guess whatever comes in front of the camera.
 8. CV - Shadow Art

Computer Vision (CV): It is a field of artificial intelligence (AI) of

machines that helps to analyse visual data (like digital images, videos and
others), derive meaningful information and patterns from them and
eventually, make decisions, take actions or give recommendations based
on that information received. Computer vision enables machines to see,
observe and understand the world.

Computer games/applications, like Shadow Art, use CV.

★ Shadow Art: It is a web browser-based game by Google that combines

AI and the art of shadow puppetry. It brings an offline experience into an
online mode with the help of Computer Vision. One can use their hands to
form shapes of the twelve zodiac animals from the lunar cycle in front of
the camera. The picture of our hands transforms into a shadow puppet
animal.
 9. AI domain - NLP

Natural Language Processing (NLP): It refers to the branch of AI that is

involved with giving computers the ability to understand spoken words (or
text) like human beings can. NLP integrates computational linguistics with
statistical, machine learning and deep learning models. Thus, computers
can process and understand the human language of text or voice data.
Applications of NLP include Google Assitant, Language Translators,
Sentiment Analysis, etc.

Computer games like Semantris and Luis.

★ Semantris: It is a word-association game that is powered by

machine learning. The AI is capable of doing so by being trained with
billions of online conversations. It predicts which words might come next
in a conversation correctly.
★ Luis: It is a machine learning-based service to build natural
language into apps, bots, and IoT devices. It helps to create enterprise-
ready, custom models that constantly improve.
 10. Teachable Machine Activity

Teachable Machine is a web-based tool that makes building machine-

learning models fast, effortless, and accessible to everyone. This AI-
based application helps us train a computer to recognize images,
audio and poses without writing a program. First, we have to gather
and group the data. Then, we have to train the model and test it out to
see whether it can correctly classify new examples. We can export
this model for our projects like websites, apps, etc.

★ Image Project: A model can be taught to classify images using

files or a webcam with at least 60 images of the object.

★ Audio Project: The model can be taught to classify audio by

recording short sound samples. At least 10 sound samples must be
recorded.

★ Pose Project: The model can be taught to classify body positions

using files or striking poses in front of the webcam.

Image:
https://teachablemachine.withgoogle.com/models/vmTiKk5w3/
 Audio:
https://teachablemachine.withgoogle.com/models/lwUpzg0Aq/

Pose: https://teachablemachine.withgoogle.com/models/easBp4en0/