SCORE :

95%
 ARTIFICIAL
Sr No 1 Marks INTELLIGENCE
Questions 2 Marks Question

SelfWanagement Skil@ll 2 SeKJvlariagement Skiltc-II 1

Ouestion Question

Digital Digital Documentation 1

Documenlation Ouestion

s No 1 Marx• 2 Marks 4 marks

euestions Question Question

Data uestions
Sčionces 1

uestions ) uation 8tion

Total Number of Questjons 24 (1Question) ’Any 4 n (1Question)
Any 20 y
INTRO TO
AI
INTRO TO
AI
INTRO TO
AI
INTRO TO
AI
INTRO TO
AI
 INTRO TO
AI

• Ensuring Al technology is available to everyone not just a

few.
QUESTIO
N

B Interpersona I
intelligence
B
Unemployment

C Ai
bias

D No
Concern
 AI PROJECT
CYCLE

The Pro ec C cle cons sts of several stages that guide the developin ificial Intelli
process ence
systems These stages ensu e a st uc u e u ding that solve real-world
intelli
 AI PROJECT
CYCLE

1 — Who are the stakeholders involved*

2 — What is the problem to be solved?
3 Where — Where does the problem
4 occur*
y y is it important to solve this
problem*
AI PROJECT
CYCLE

e.
M,
 AI PROJECT
CYCLE

• Trainin : Used to teach the model how to recognize

• Testi patterns ata.sed to evaluate the model's
performance
 AI PROJECT
CYCLE
Gata Sou
rccs:
1. S i ollecting user opinions through forms and
‘°'" questionnaires
tomer satisfaction surveys for product
2. improvement
xam era .ii‹rt. xtracting data from websites using
7.’e automated tools. ecting news articles to train a
3 sentiment analysis model
‹‘
xample. evices that
Collecting gatherdata
weather datausing
from loT
thesensors
environment
4. a (temperature.
Visua motion. etc.) ou h video
capture
5. Example.
'›r. . a ttcn-s T.raffi
c monitoring
Manual or throug
uto ated eco cameras
d g of data by observing a procsys
pieE a racking customer movement patterss i t a o r e
6 s : I.caticn F’rogrrJf1Jilt!r‹:I IntOlfascs Accessing data from
. external applications.
app e. Using Google Maps API for real-time traffic data
AI PROJECT
CYCLE
 AI PROJECT
CYCLE

3 E A scatter plot with a third dimension represented by bubble

size
Example. Sales revenue. profit margin. and product categories
 AI PROJECT
CYCLE

AI systems that follow predefined rules to make

decisions

• Interpretability. Easy to understand and explain

• Transparency. Clear decision making process
• Ease of Maintenance Simple to update rules
• Limited Expressiveness Cannot handle
complex data
• Scalability: Difficult to scale for large datasets
 AI PROJECT
CYCLE

¥
O
Example Re positive prædictions out of actual positives
• F1 Score: Harmonic mean of precision and recall

Example. Evaluating a medical diagnosis Al by checking its accuracy in predicting

diseases
 AI PROJECT
CYCLE

eural networks mimic the human brain o atterns and

iden i decisions. ey consis oetc e co make f neur ns
ed la ers

1. !: :u: Accepts raw data for processing °

Example Image pixels in an image recognition system

2. i d- L Perform featur extraction and atlern

learning
Example. Identifying edges and shapes in an image

'ñ\itg t L ‘ier: roduces the final prediction or decision.

Example ecognizing a handwritten digit
 AI PROJECT
CYCLE

• Non linea Ca tures com lex relationships in data

• Ada a iliLearns rom arge a ase s and adapts to
• Generalization:changes.
Can make accurate predictions on unseen
data
• Parallel Processing: Processes multiple inputs
Ex ample. Facial recognition systems use neural networks to identify people in images ducing image resolution
simultaneously
while
preserving key features

3. Reinforcement Learning: he model learns by interactin with the environment

and eceiving rewards
Example Teaching a robot to navigate a maze through trial and error
AI PROJECT
CYCLE
QUESTIO
N

C Decision
tree

D K Map
QUESTIO
N

C Cloud
Computing

o None of
These
QUESTIO
N

o Web
scraping
QUESTIO
N

learning
models
QUESTIO
N

A Learnin -Based A r h in AI refers to a system that and akey decisions from L

Senaosns atte doata stgea r eo n re e and Lea ni (DL)
ed rues It uses Machine Learnin techniques to improve its
performance over time based on experience

Unlike Rule-Based Approaches. learning-based systems adapt to new data. making them
more flexible and effective for complex and dynamic problems Iike image recognition,
natural language processing. and recommendation systems
4o
• Data Science includes multiple domains such as
Statistics: Analyzing and inter etin data
patterns (e.g
calculating averages to identify trends)
nn Building models that learn w.
predictions (e.g from data email
. spam to make
detection)
a n nee n siotonnd a
ed (e.g building data pipelines)
Visua atio Presenting data through charts
i and graphs (e.g
et findings to . pie charts for market share) ""“”"
Bi Handling massive datasets that are too ”
large
tra for i ional processing (e g . analyzing social media
trends) Busi Using data to support decision-making
(e.g.. tracking customer behavior)
 DATA SCIENCE

Fraud detection. (e g. predicting loan defaults. detecting credit card

• rig
f,J .i :°’ iiCeasrto fraud)
. re at on s stems (e g personalized product
recommendations,
arge e a venising)
• orc emand forecasting. customer behavior prediction (e.g.. Amazon product recommendations.
• stock management)
.‹a'.. .s daptive learning. performance prediction (e g. personalized tutoring systems. analyzing
• student progress e
iie r:aic!; er» Content recommendations, user behavior analysis
(e.g..
watch history)
 MOnJtortng Systems Ç
Manufacturing

Main lenznce Scf+edul

ng

Transport

Sinking
 DATA SCIENCE

2. Data Ac:quisitio!
i
• Ga herin and relevant data for model traini
collec in custesting
o e a mmendation system or weather data for climate
• a p es models
Collectin
• Understanding and analy llected data through visualization and statistical techniques
• Examples Using histograms to understand age distribution or scatter plots to identify relationships
between variables
 DATA SCIENCE

Buildin a e mach ne earning algorithms.

• Example Training a classification model to identify s p a da es

• Assessin the odel'sice sn trics such as accuracy. precision. recall. and

F1-score
• Example Evaluating a medical diagnosis model's accuracy in identifying diseases

”' e|: C'\’iz e ° aic ^C Crl t:*a‹:I:

• Deploying the model for real-world use and gathering feedback for improvements
• Example: Implementing a chatbot for customer service and refining it based on user interactions
 DATA SCIENCE

• Data
Collection

• use s (e.g. feedback forms from customers)

• min (e g collecting product prices from e-commerce
• websites)
T
• D
(e g . monitoring temperature in smart homes)
(e.g.. government census data)
• u ic a asets (e.g. Kaggle. UCI Machine Learning
Repository)
DATA SCIENCE
 DATA SCIENCE

• Structured: Data organized in a tabular form (e.g.. databases. Excel

sheets)
• Unstructured. Data without a predefined format (e.g. images. videos.
audio)
• Semi-structured: Data with some level of organlz ation (e.g.. JSON. XML
files)
• CSV (Comma-Separated
Values)
• JSON (JavaScript Object
Notation)
• XML (eXtensible Markup
Language)
• Excel (.xlsx)
• SQL databases
 DATA SCIENCE

• fast and effi i n m I i imen i n I array for data

manipulation
• Examp e toring pixe va ues or image analysis

• Series: 1D labeled array (e.g . a column oí student marks)

• DataFrame: 2D labeled data structure (e g. a table with multiple columns and
rows)
 DATA SCIENCE

• owa na anipu a a d clean

• teyration with other libraries like Numpy and

Matplotlib
 DATA SCIENCE

• }§§ jg„ tting options (line plots. bar charts. etc.)

• Customizable charts (labels. colors. markers)
• n eg a on with Numpy and Pandas for seamless data
handling
 DATA SCIENCE

• verage value (e.g.. average exam score)

• iddle value in sorted data (e.g. median
income)
Mode: Most frequent value (e.g.. most common customer age)
am ard Deviation Measures data spread (e g how much students'
scores vary) Measures e.g. assessing data consistency)
• rr I ion Measures the relationship s and exam scores)
Re betwee
ression dentifies the relationship betwee ependen riables (e.g. predicting
an independent prices base on square footage) house
 DATA SCIENCE

• Es sential for understanding data patterns and model performance

C on.mon T'/t›cs of Da ta ü ollec tion Probicn1s

• n om lete dat issi values (e.g. incomplete survey

responses)
• Inconsistent data Mismatched formats (e.g., date inconsistencies)
• Biased data Skewed representation (e gonly including certain
demographics)
 DATA SCIENCE

Ii'

Trends (e.g. temperature changes over

o months)
betware
Proportional data (e (e.g sales by
or es distribution)
g.. market share
• product) (e g . age vsincome)
Relationships between
• s Data distribution (e.g.. exam score frequency)
 DATA SCIENCE

• Analyzing data to classify and predict personality traits

• Example Predicting personality types using social media
behavior

• A simple. non-parametric algorithm used for

classification and regression

1 Stores all training instances

2. Calculates the distance between the test point and
all training points.
3. Selects the 'k' nearest neighbors
4. Predicts the class by majority vote
 DATA SCIENCE

• Simple and easy to implement

• No training phase required (lazy
learning)
• Works well with small datasets

• Computationally expensive for large datasets

• Sensitive to the choice of 'k'
• Requires feature scaling for better performance (e.g.. normalizing age and income
together)
QUESTIO
N

A. Median

B. Mean

C. Standard
deviation
QUESTIO
N

B
Pandas
QUESTIO
N

In the Al domain of Data Science. data refers to the raw information

collected
from various sources. which is then processed and analyzed to extract
meaningful insights Data can be of different types including

••••••p 1.Structured Data — Organized in tables (e g.. databases spreadsheets)

2. Unstructured Data — Includes text images. videos. and audio
3. Semi-structured Data — A mix of structured and unstructured
data (e.g. JSON. XML files)
COMPUTER
VISION
 COMPUTER
VISION

!i r‘T L›u . :‹ are Enabling vehicles Io detect and understand their environment.
including road signs pedestrians and oher vehicles
ii !.'ec.°.a. i!: at? ?na ',‘sis Assisting in the diagnosis of diseases by analysing X-rays. MRls
and other medic£ll images for abnormalities

!t: ¿/ ciecï ioo Recognising and locating objects

within images or videos. useful in retail for inventory management and surveillance

ir ‹: ?r.r. '.<.'ldc Monitoring and analysing video feeds for security purposes. detecting unusual
activities or behaviours

v it c .? ssi! < ion Categorising images into predefined classes for applications like
photo organisation and content moderation— ’
 COMPUTER
VISION
Computer Vision Tasks Computer vision applications rely on several key tasks to extract information trom
images foror further analysis
prediction \
6^
j
i ‹? css.‘ , input image from a fixed set of categories. forming a core
' *QThis task ass gns a labetto ad’
problem in
computer vision with many practical applications
!<.ss c .io. I ‹° :r. iset an T,his task involves identifying both what object is in the image
and its specific location. typically used for single objects

‹ ñ;':‹:l L tñ o QThis process finds instances of real-world objects. like faces or bicycles. in
images or videos. often using algorithms that cognise object categories It's commonly applied in image
retrieval and automated parking systems
¿st¿. .+n This task detects individuaobjec y'ate ses m. and,labeIs eachi I
t/¿ in segmented regions of the
resulting accordingly.
image
 COMPUTER
VISION

We encounter images daily through our phones and computers. but understanding their fundamentals is
important
is ics cf F’.• pixel. short for "picture element” is the smallest unit of a digital image. typically
›s arranged in a result in a ecarer image. atsheyclosely approximate the original picture

* J‹ i Resolution refers to the number of pixels in an image. often expressed as width by height
1 80 (e.g means there are 1280 pixels horizontally and v‹ '
1024 .
1024 venically It can also be represented as a single number. such as 5 megapixels (5 million pixels)

›.›/ "›:‹ci .›!'›.c E,ach pixel has a value that indicates its brightness and color. commonly stored as an 8-bit
integer anging rod (black) to 255 (white). This range is due to the binary system. where 1 byte equals 8
bits allowing for
 COMPUTER
VISION

F!‹ ’ ›‹.‹'c Coloured images are made from three primary colors Red. Green and Blue. Different
intensities of
these colours combine to create the wide range of colors we see in images
 COMPUTER
VISION

In computer vision and image processing, a feature is a relevant piece of information used to solve
specific
computational tasks Features can include structures within an image. such as points. edges. or
objects
QUESTIO
N

A Price
PríÓiction

D Face Lock in Smart Phone

QUESTIO
N

C Wee nd e
s

D High Resolution
Images
QUESTIO
N

C Long Short- Term Memory

(LSTM)

D Hopfield Network
QUESTIO
N

B. 0 to I

C. —128to
T27

D 1 tO
100
 NL
P

(ii) a n co uttrs h a n a u n t s o f
NLP makin te t nd eas te h mach nes
g
 NL
P

ż>ont i n ys s nalysing text to determine the sentiment such as

i behind it positive
neg ions
a.\ ci i!-.. Inst°.I r› Translating text from one language to another Iike Google
Translate

Condensing long articles or documents into shorter summar Yes while

rela n ng essen a n o mation
':co° I’ocooi on onverting spoken language into text. used in voice-controlled
devices and transcription services.
 NL
P

(i) A chatbot is a computer program designed to simulate It can

answer
uestionoswn ovide noa an ass s as s through text or voice
interactions
 NL
P

.”I'.i?? i0 I .:—L'NJ?C

( i ) Iaru a a una es i e English. Spanish or Mandarin that people use to

communicate

(ii) Rich in nuance, context and emotion. making it complex and flexible

(iv) e a b uous and relies on tone. culture and shared experiences

 NL
P

1 Humans communicate easily using natural languages. but these languages are
complex
c for rstand
2. g|$[g| [g§§§§|[]g }@,P) helps machines interpret and respond in human
languages
Egg by converting text into numerical data
3 This process starts Cth Text Normalisation. which simplifies text to
make it more understandable.
 NL
P

".oo »‹ too.. or«s ?*'ccc.al I s Unnecessary tokens, like common words (e.g.
arac.c s aod I? r: «L ”is"). are removed to focus on ”the".
meaningful content
:c.n educes words to their root forms by removing prefixes and suWixes. For example.
"healed" "healing". and "he9ler" become "heal"but "studies" becomes "studi", which may not be
meaningful

La '« «aiis:°.!‹Ji' Similar to stemming, but ensures the resulting words (lemmas) are meaningful. This
process takes more time than stemming

?a r of ‘. .oics A model that counts the occurrences of each word in the text to create a vocabulary.
which is
usefu or machine earning tasks
 NL
P

i. ?“ci“.c e¿ ›=i .in Breaks the text (or corpus) into individual sentences for easier processing

ii. c« s : Further divides each sentence into smaller units called tokens. which can be words.
specialo characters.
numbers or

"is") are removed to focus on meaningful content

S: « Reduces words to their root forms by removing prefixes and suWixes. For example.
"healed"
"healing". and "healer" become "heal" but "studies" becomes "studi", which may not be meaningful

L « «*iis:°.! oi‘ Similar to stemming, but ensures the resulting words (lemmas) are meaningful.
This process
takes more time than stemming

?a of ‘. oics A model that counts the occurrences of each word in the text to create a vocabulary.
which is
useful (or machine learning tasks
QUESTIO
N

C Named Entity Recognition

(NER)

D Machine
Translation
QUESTIO
N

B To convert text into lower

case
QUESTIO
N

D Detecting emotions in
images
QUESTIO
N

C Translates text from one language to

another

o Analyzes the structure of

sentences
 Evaluatio
n

2. It's important not to use the training data for evaluation. as this can lead to overfitting. wnere
the model
memorises the training data instead of generalising from it
 Evaluatio
n

While it itself is not an evaluation metric. it serves as a useful tool for assessing prediction
accuracy.

Now that we understand the various outcomes of predictions versus reality. let's explore how to
evaluate the

g p ti atio A d correct if
matc u o t it
h e
 Evaluatio
n

Precision is the percenta e of true positive predictions compared to all predicted positives. It focuses
on True Pos ves an a r s e Positives

Recall indicates the fraction of actual positive cases that were corre,ctly identifid It primarily
considers
 Evaluatio
n

Precision is the percentage of true positive predictions compared to all predicted positives. It focuses
on True Positives and False Positives

recognised
it
QUESTIO
N

C Recall and Accuracy

o Recall and
reality
QUESTIO
N

C
Reliability

D Punctualit
y
QUESTIO
N

C
Accuracy
INTRO TO
AI
Topic: Homework From ACE
Module
SCORE :
95%
QUESTIO
N

A
Abrupt

B Accountabl
e

Achievable

D
Admirable
QUESTIO
N

A Prioritizing ’/our
work

Ț Not taking

teedback C Goal setting

D Staying updated about new

practices
QUESTIO
N

ø Hafdvvofing

B Resistance to
change

LOZ'y'

o Less—confident
QUESTIO
N

A Web Browsers

Operating
System

C Office Software

o Designing
Software
QUESTIO
N

A. Written
communication

B. Public
communication
C Small QrOUD
COmmunication

D Interpersonal
communication
QUESTIO
N

Both ( ) and (R) or=. true and (R) is the correct explanation for
( ).

B Both (A) and (R) are true and (R) is not the correct explanation
of (A).

C ( ) is true, but (R) is

false.

o ( ) is false, but (R) is

true.
QUESTIO
N

A High Model Complexity

P Relevant and Authentic Traininn

*n

C Minima I
Preprocessing

o Limited Hardware
Resources
HQ. This real-Iif- nnn!ircition ot ITU is used to provide an o«erview ot o ne*z s
item Dloo
or post z.hile civoidina reduncioncv from multiple s 'arc onO
moximiz»g the di* rsity of content obtained. \Vhich is this application? (!
nn)

A. Chatbot

B. Virtual Assistant

C. Sentiment Analysis

Fy Automatic Summarisation
QUESTIO
N

A A robotic vacuum cleaner that can navigate and clean floor* autonomously.

B A chatbot that enannes in natural language conversations and answers q.

estions.

C A smartphone with! facia I reconnition f or unlocking the

device.

^ /. digital alarm ciock that rings at a set time every

morning.
QUESTIO
N

/. Care

B. Cared

C. Care
s

D. Car
QUESTIO
N

A Measures an individual s ability to understand: o‹' ers emotions and

feelings.

Assesses one's proficiency in mathematics and logical reasoning.

Z- Describes the level of self—awareness someone has, startinq from

realizing weaknesses, strengths, to recognizing their own feelings.

Ș/ EVoluates an indiViduaI’•. spotia I navigation and

visualization skills.
QUESTIO
N

A CSV

Website

C SQL

o
Spreadsheet
QUESTIO
N

A
Precision

B Accurac
y

•“,
Recall

D F1
QUESTIO
N
! C IN .I"ü .'l ? C P S U ' üiC i OT i'›ï iOlOY i.^C .G? S UTC : OI¢Ci Oi S ? J S °
iü|?? l,Se’ CA

A Data Exploration, Problem Scoping, Modelling, Evaluatior. Data Ac ,ui

ition.

P Problem Scopinc, Data Acquisition, Data Exploration, Modelling,

Evaluation.

C Modelling, Data Acquisition, Evaluation, Problem Scoping, Data tion

Exp! .

D Data Acquisition, Data Exploration, Problem Scoping, Modelling,

E'valuation.
 is a concept to unify statistic* data anale sis, machine learning and
their

A Computer
Vision

B Natural •nca<sin
Lanniinn n

Data
Science

D Computer
Science
QUESTIO
N

A. Classification

B. Classification +
Localisation

f• Instance I egmentation

D Localisatio
n
QUESTIO
N

A When a legitimate email is accurately identified as not

spam.

Ș When c. spam email is mistakenly identified as

iegitimate.

C When an email is accurately recognized as

spam.

o None of
these
HQ. 4D Gulati stcii tech‹ithc small shop 'ith his focus, dedication ance clear icJeas,
MDH Reccme one of the most popular brands in Inclic, besides hcvii g o good
reputation all ovct the •/voi“lcJ".
*A*nich self- management ski I is clearly visible in the given sta temcnt? (1ł4)
QUESTIO
N

A. DrOg
onddrop

B. Double click

î” Hover

D Single
click
QUESTIO
N

A. Both ( ) and (R) are correct and (R) is the correct explanation
of ( ).

B. Both ( ) and (R) are correct Dut (R} is NOT the correct explanation

of (A). Ț ( ) is correct but (R) is not correct.

o ( ) is not correct but (R) is

correct.
#Q the ork is ali about iden*ifyin and how ›//e soent out time
notin onIl/ no how to spend our time and
effectively. (1tvl)

A Organizin
g

B. Prioritizing

C. Controllin
g

/•y Tracking
QUESTIO
N

A Interpersonal
barrier
B Physical
barrier

C Organizational
barrier

I Linguistic
° barrier
QUESTIO
N

Ț. Both ( ) and (R) are correct and (R) is the correct explanation
of ( )
B Both (A) and (R) are correct Dut (R} is not the correct explanation
of (A).

C ( ) is correct but (R) is not

correct.

o ( ) is not correct but (R) is

correct.
HQ. The Indian Gov'ernment baiuaer a few a o|3S StCJtina, "Ser'vers in the hostile
are recei»inu and using the acquired dota
rzation
improperly".
\^/hicln terimiiaolog / suits best tor this action? (!lvt}

A AI Ethics

Data
Privacy

C Al
Bits

o Al
Access
QUESTIO
N

A. Both Statement I and Statement 2 are c-.

ect.

B. Both Statement 1 and Statement 2 a°.'

incorrect.

W Statement 1is correct but Statement 2 is

incorrect.
D Statement 2 is correct but Statement 1is
incorrect.
QUESTIO
N

A iPs

B
doc
É" CS
Ș V
QUESTIO
N

A. Stop word

B. Rare word

C. Frequent
word

o Removable word
HQ. Read the examples given belo : (1łvI)

Fare ‹inlock techno!oqy of mobile Dhon s iisinc camera

T«nninc orf lie its faith mn InT wire
i V. Hand sanitizer dispenses having se asor Choose the options that are
not AI.

A. i and ii

B. iii and i

iii and iv

o , iii and
iv
QUESTIO
N

A Text summa risation

P' Target
adVertisements

C Face lock in
smartphones

D Email
filters
QUESTIO
N

Ț, Both A and R are correct and R is the correct explanation of

A.

B Both A and R are correct but R is not the correct

explanation of A.
C A is correct but R is not
correct.

o A is not correct but R is

correct.
QUESTIO
N

A. A. Access

B. data
Privacy

C Al Bits

AI
Ethics
QUESTIO
N

A. Both Statement I and Statement 2 are

correct.

B. Both Statement ” u d Statement 2 are

incorrect.

Statement 1is correct but Statement 2 is

D incorrect.
Statement 2 is correct but Statement 1is
incorrect.
 file form°.cit is common lv used for s*orinq and exchanging doto
stsuctui ed,inPetula
ci r form?

(1M)

A
mp3
B pd
f

f"
csv

o gif
QUESTIO
N
lO ? ÇUSGI
CO l

A. Stop word

B. Rare word

C. Frequent
word

o Removable word
 A otnotocooy mock it•.e G' ’Ș!'ostinrJ
documents Gooc le Translote transloiing
lanquacies instoutly h ».oshinq mcichine with
a pre—sent timer
v. A chutUot ao.szvering customer questions

A i, ii, and iii

^ ii and
iv

C , iii, and
iv

D i and
iii
QUESTIO
N

@ Weather
forecasting

B Spell checking in
documents

C Playing video
gomes

D Formatting text in a
document
QUESTIO
N

(J
M)

A. Instance
segmentation

B. Object

detection

D Image
Classification
segmentation
QUESTIO
N

A High Precision and low

Recall

B Low Precision and high

Recall

A good balance between Precision and

RecaII

o Only good accuracy

QUESTIO
N

A. Both Statement I and Stat ment 2 are

correct.

B. Both Statem•..ot 1 and Statement 2 are

incorrect.

Statement 1is correct but Statement 2 is

D
incorrect.
Statement 2 is correct but Statement 1is
incorrect.
QUESTIO
N

A. To improve the model's ef(iciency

B. To check the accuracy of predictions

C To clean the data

D To collect more data for

training
QUESTIO
N

A Sentiment
Analysis

z* Text Classification

C Machine
Translation
D summarizatio
n
HQ. Meena created a model to classify plants by using data from
her local garden
\'\*hen tested on plants front different regions, it couldn t classify them accurotel\•.
\^/!aot is the ssue heie? (1M)
QUESTIO
N

A. Data Science

B. Robotics

Computer
Vision

D Natural Language
Processing
QUESTIO
N

A. colour

B. monochrome

P grayscale

D
bitmap
QUESTIO
N
(1
M)
HQ. John » as feeling stressed out at ø ork. Üe tnaa a lot of
ueaČlines to meet,c demanding toss cnci a grow ing worklOod. He ?
ne•N he needeci to do soivetnii ¿ to mcinc qe his stress levels alab
improve his o ercill well-being
*A*hat shouted John do to mancige his siress levels? (IV)
QUESTIO
N

A Start
Menu
B System
Tray

C. Desktop
lcons

D. Taskbar
QUESTIO
N

(1
)

A Both A and R are correct and R is the correct explanation

of A.

B Both A and R are correct but R is NOT the correct explanation

of A.

C A is correct but R is not

correct.

o A is not correct but R is

correct.
QUESTIO
N

A. physical

B. emotiona
l

C. mental

D tinancia
l
QUESTIO
N

(1M
)
A. Courtesy

B. Conciseness

C. Concretenes
s

D CGłnsistency
QUESTIO
N

A
17

B 1
5

C
13

D
19
QUESTIO
N

A. Both A and R are correct and R is the correct explanation of

A.

B. Both A and R are correct but R is not the correct

explanation of A.
C A is correct but R is not
correct.

o A is not correct but R is

correct.
QUESTIO
N

A AI
Access

B AI
Ethics

C Al Bits

D Data
Privacy
QUESTIO
N

A. Both Statement I and Statement 2 are

correct.

B. Both Statement 1 and Statement 2 are incorrect.

C Statement 1is correct but Statement 2 is

incorrect.

D Statement 2 is correct but Statement 1is

incorrect.
QUESTIO
N

A. mp
3

B. P^!
C CS
V

o gif
QUESTIO
N

A. Stop word

B. Rare word

C. Frequent
word

o Removable word
HQ. *›’7hicln outions do NOT involve (IV)
Al? Text gi edic tion while tvpint an
emaii Speedometer in ci car
displaying speed
Pe sonal c'ssistcnts like Siri pio‘ idinc
recommenclations
iv. Digital clock displciying the time

A ii and iv
B i and
iii

C i.e., iii. and

iv.

D i, ii, and
iv
QUESTIO
N
?'ï t7! ¢I 'CCL lOlOï' i? LS?C - CCü|??!C Y?C??Si'?S tO lüCOilT’??OC. JtO?ü:?tS ?
^C ö/ G

A. Speech recognition

B. Product recommendation
systems

C mage
filtering

D
CAPTCHA
QUESTIO
N

A. Instance
segmentation

B. Object detection

C. Classification

D Semantic
segmentation
QUESTIO
N

A TP/(TP +
FN)

B (TP + TN)/(TP + TN+FP+ FN)

C TP/(TP + FP)

D TP/(TP + FN +
FP)
QUESTIO
N

A. Both Statement I and Statement 2 are

correct.

B. Both Statement 1 and Statement 2 are incorrect.

C Statement 1is correct but Statement 2 is

incorrect.

D Statement 2 is correct but Statement 1is

incorrect.
QUESTIO
N

A Testing
Data

B So mple
Data

C Training Data

D E\valuation Data
HQ. '›’7hot does CV stond fot in AI c1nd image (IM
processing? )
QUESTIO
N

(J
M)

A. Instance
segmentation

B. Object detection

C. Classification

D Image
segmentation
HQ. Rcivi trained a claatloot using con /ersotions ioet«een people ›r/ho cill had a
similar oomicuinicotion style. \A he nee/ users tried th chatbot, it couIcln't
understated their
^,'o‘/ of s¡oeaking. \^/taat concert is being illustrated here? (1M)
QUESTIO
N

A. Data Science

B. Natural Language
Processing

C. Computer Vision

D. Robotics
QUESTIO
N

A. Pixel

B. Bit
Depth

c lntensity

o
Resolution
QUESTIO
N

A. Human language has fewer rules.

B. Human language is less expressive than computer

language.

C Human language can be ambiguous and context-

dependent.

o Computer language requires interpretation by

humans.
QUESTIO
N

A To develop
robots
B To understand and generate human language

C To create computer
graphics

D To optimize
aIgorithms
QUESTIO
N

A. Fraud detection

B. Disease outbreak detection

C. Customer sentiment
analysis

D Credit score
prediction
QUESTIO
N

A Identity
theft
B Cyber
theft

C E—commerce
fraud
QUESTIO
N

A. Moving

B. Renaming

C. Copying

D Deletin
g
QUESTIO
N

A. Both A and R are correct and R is the correct explanation of

A.

B. Both A and R are correct but R is not the correct

explanation of A.
C A is correct but R is not
correct.

o A is not correct but R is

correct.
QUESTIO
N

A
psychological

B physiologica
l

C both of
them

D none of
these
QUESTIO
N

A. Encoding

B. Decoding

C. Feedback

D. Transmissio
n
QUESTIO
N

A. Clean Water and Sanitation

B. Gender Equality

C. Population

D. Reduced Inequalities
QUESTIO
N

A Both A and R are correct and R is the correct

explanation of A.
B Both A and R are correct but SRL is not the correct explanation
of A.

C. A is correct but R is not

correct.

D. A is not correct but R is

correct.
QUESTIO
N

A Data
Privacy

B AI
Biøs

C. AI Ethics

D. AI
Access
QUESTIO
N

A. Both Statement 1 and Statement 2 are

correct.

B. Both Statement 1 and Statement 2 are

incorrect.

C. Statement 1 is correct but Statement 2 is

incorrect.

D. Statement 2 is correct but Statement 1is

incorrect.
HQ. Coi°.aoIete tm
i e sentence based on the graph: (J M)
The Ic r¿er the size of the ticoining dcitciset, the __ _ _ __ _ (betters ^.orse) the
model's
generalization aIoiIit\
ïmpoct of Tra ining Ooto set Size on Test Aecu roe
y

8 .
6

#
70

Tra ining D a t a set

Size
QUESTIO
N

A. Storing audio files

B. Processing online transactions

C. Storing and analysing large volumes of

data

D. Compressing images
QUESTIO
N

A. Stop word

B. Rare word

C. Frequent word

D. Removable word
QUESTIO
N

A ii and
iii

B l and
iv

C ii, iii, and

iv

D i,ii, and
iv
QUESTIO
N

A. Stop word

B. Rare word

C. Frequent word

D. Removable word
QUESTIO
N

A Video
editing

B Disease
diagnosis

C. Weather forecasting

D. Currency
conversion
QUESTIO
N

A. Classification

B. Object detection

C. Image
segmentation

D. Feature extraction
QUESTIO
N

A. True Positive

B. Fake Positive

C. True Negative

D. False
Negative
QUESTIO
N

A Video
editing

B Disease
diagnosis

C. Weather forecasting

D. Currency
conversion
QUESTIO
N

A. Both Statement 1 and Statement 2 are

correct.

B. Both Statement 1 and Statement 2 are

incorrect.

C. Statement 1 is correct but Statement 2 is

incorrect.

D. Statement 2 is correct but Statement 1is

incorrect.
QUESTIO
N

A. To finalise the project goal

B. To collect and understand the data features relevant to the

problem

C. To analyse the system map for unrelated data

D. To test the AI model without data

QUESTIO
N

A Tokenisation

B Parsin
g

C. Summarisation

D. Sentiment
Analysis
QUESTIO
N

A Natural Language
Processing

B Machine
Learning

C. Robotics

D. Data
Science
QUESTIO
N

A To process
images

B To analyse data
patterns

C. To understand and generate human language

D. To build physicsI robots

QUESTIO
N

A. Supervised

B. Unsupervise
d

C. Reinforcemen
t

D. All of these
QUESTIO
N

A The model is too

strict

B The model incorrectly pedicle positive

outcomes

C. The model has low accuracy

D. The model is perfectly

calibrated
QUESTIO
N

A The model is too

strict

B The model incorrectly pedicle positive

outcomes

C. The model has low accuracy

D. The model is perfectly

calibrated
QUESTIO
N

A. Formatting

B. Compressin
g

C. Encrypting

D. Copying
QUESTIO
N

A. Both A and R are correct and R is the correct explanation

of A.

B. Both A and R are correct but R is not the correct

explanation of A.

C. A is correct but R is not correct.

D. A is not correct but R is correct.

QUESTIO
N

A. Acute stress

B. Chronic
stress

C. Episodic
stress

D. Physical
stress
QUESTIO
N

 Summative
feedback

B Descriptive
feedback
C Genera I
feeÖDack
D Evaluative
feedback
QUESTIO
N

A. Clean Water and Sanitation

B. Gender Equality

C. Population

D. Reduced Inequalities
QUESTIO
N

A. Both A and R are correct and R is the correct explanation

of A.

B. Both A and R are correct but R is not the correct

explanation of A.

C. A is correct but R is not correct.

D. A is not correct but R is correct.

QUESTIO
N

A. AI
Access

B. AI Ethics

C AI
Bias

D Data
Privacy
QUESTIO
N

A. Both Statement 1 and Statement 2 are

correct.

B. Both Statement 1 and Statement 2 are

incorrect.

C. Statement 1 is correct but Statement 2 is

incorrect.

D. Statement 2 is correct but Statement 1is

incorrect.
HQ. Fill in the blank: (J
M) As dropout rate increases, the model's tencienc\ to _ _ _ _ _ _ _ _ (overiit ,*
under fit)
reef uces.
Effect of Dropout date on
OvcNitting
QUESTIO
N

A
json
B xml

txt

D
excel
QUESTIO
N

A. Stop word

B. Rare word

C. Frequent word

D. Removable word
HQ. \\'hicia of these cipulications does NOT utilise
Al?
ii. Spotify su$gesting songs hosed on }•our listening
histor/
iii. Self-iiivin ccirs using sensors for
navigation
iv.f•Jicrow’ove heating food »vitIn a present
timer Choose the o¡otions thai are not AI.
(1M)

A. only i

B. iii and i

C. iii and iv

D. i, iii and iv
#Q is lie branch ot Al tnat deals with enabling machines to undei
stand and genei cte i ru an an quote. (1Ivf)
QUESTIO
N

A Fraud
detection

B Autonomous
driving

C. Climate
forecasting

D. Voice assistants
QUESTIO
N

A. Classification

B. Object detection

C. Image
segmentation

D. Image filtering
QUESTIO
N

A. Precisio
n

B. Recall

C Specificit
y

D
Accuracy
HQ. Arjun needs to iclentify the AI techniquethat focuses
on giving comautc's th ability to ii tergret ence understand
visual informa:ion froth ti e .^ orId. 'A'hct is tt is ficicJ ca led? (1M)
QUESTIO
N

A. Both Statement 1 and Statement 2 are

correct.

B. Both Statement 1 and Statement 2 are

incorrect.

C. Statement 1 is correct but Statement 2 is

incorrect.

D. Statement 2 is correct but Statement 1is

incorrect.
QUESTIO
N

A. Data
Validation

B. Data
Acquisition
C Data
Cleaning

D Data
Evaluation
QUESTIO
N

A Sentiment
Analysis

B Speech
Recognition

C. Text
Classification

D. Entity Extraction
QUESTIO
N

A Data
Science

B Machine
Learning

C. Natural Language Processing

D. Computer Vision
QUESTIO
N

A Colour Depth

B Pixel
Count

C. Resolutio
n

D. Bit Pate
QUESTIO
N

A Natural Language
Processing

B Neural Logic
Programming

C. Nonlinear Programming

D. Numerical Language
Parsing
QUESTIO
N

A Word
embedding

B Term frequency-inverse document frequency (TF-

IDF)

C Bag of words

D Part of speech
tagging
QUESTIO
N

A A spam email marked as not

spam
B A legitimate email markedas
spam

C. No emails detected

D. All spam emails

identified
deme nding ‘moss and a gi oWiI g workload. He knew lie need d to do
something to manage Iris stress Ie»cIs and improve his overall ‘veil—being.
\^.'hat do v°ve call good stress?(J M)
QUESTIO
N

A. Moving

B. Renamin
g

C. Copying

D. Deleting
QUESTIO
N

A. Both A and R are correct and SRS is the correct explanation

of A.

B. Both A and R are correct but R is not the correct

explanation of A.

C. A is correct but R is not correct.

D. A is not correct but R is correct.

QUESTIO
N

A. Emotional
stress

B. Physical stress

C. Mental stress

D. Social stress
QUESTIO
N

A Semantic
Barriers
B Physical
Barriers
C. Psychological
Barriers

D. Organisational
Barriers
QUESTIO
N

A No
Poverty

B Zero
Hunger

C Climate
Action
D Artificia I Language
Processing
QUESTIO
N

A Both A and R are correct and SRS is the correct explanation

of A.

B Both A and R are correct but SRL is not the correct explanation
of A.

C A is correct but R is not

correct.
D A is not correct but SRL is
correct.
QUESTIO
N

A. AI Bias

B. AI Access

C. AI Ethics

D. Data
Privacy
QUESTIO
N

A. Both Statement 1 and Statement 2 are

correct.

B. Both Statement 1 and Statement 2 are

incorrect.

C. Statement 1 is correct but Statement 2 is

incorrect.

D. Statement 2 is correct but Statement 1is

incorrect.
QUESTIO
N

A. Unlabeled data

B. Labelled data

C. Compressed data

D. Raw data
QUESTIO
N

A. Stop word

B. Rare word

C. Frequent word

D. Removable word
QUESTIO
N

il \e:i| recG':rne:Jngna eïous/eoic

esDosedonl

A i and
iii

B i, iii and
iv

C. ii and iv

D. ii,iii and
iv
QUESTIO
N

A. Neural networks

B. Speech
recognition

C. Image processing

D. Machine
translation
Topic:
Summary
Topic: Homework From ACE
Module
AI 4J7 CASE
BASED
QUESTIO
N

A Interpersonal
barrier
B Physical
barrier

C Organizational
barrier

o Linguistic
barrier
QUESTIO
N

A Meaning of a
sentence
B Grammatical structure of a
sentence

C Semantics of a
sentence

D Synonym of a sentence
QUESTIO
N

A
Descriptive

B Verbal

C
Spoken

o Non—
Verbal
B. Cultural
barriers

C. Gender
barriers
D Attitudinal
barriers
QUESTIO
N

A. Principle of clarity

B. Principle of timeliness

C. Principle of
consistency

o Principle ot
feedback
QUESTIO
N

C Spoken
words

D Facial
expressions
QUESTIO
N

A. Feedback

B. Encoding

C. Quarantin
e

D Decoding
QUESTIO
N

A
Visual

B Verbal

C Non-
verbal

D Writte
n
QUESTIO
N

A
Descriptive

B Verbal

C
Spoken

o Non—
Verbal
QUESTIO
N

A Language
barriers

B Cultur al
barriers

C Gender
barriers

D Attitudinal
barriers
QUESTIO
N

A. Principle of clarity

B. Principle of timeliness

C. Principle of
consistency

o Principle ot
feedback
QUESTIO
N

A. Hand
gestures

B. Signals

C. Spoken
words
D Facial
expressions
QUESTIO
N

A. Visua I communication

B. Oral communication

C. Written
communication

o Gestures
communication
QUESTIO
N

A. Feedback

B. Encoding

C. Quarantin
e

D Decoding
QUESTIO
N

XO! ?(GB Oi OSS? LOi.(

C /
QUESTIO
N

XO! ?(GB Oi OSS? LOi.(

C /
QUESTIO
N

XO! ?(GB Oi OSS? LOi.(

C /
QUESTIO
N

XO! ?(GB Oi OSS? LOi.(

C /

c'on(/: enaicie łGc'cc':scar':/cncí s.ooöcessseciiccecs

QUESTIO
N

XO! ?(GB Oi OSS? LOi.(

C /
HQ. Fill in the Blcinks

_ feedbac k in›. olves detailed comments or

conversations.

2. Non-versaI communication i cIua'es _ _ _ _ _ _ _ expressions, gestui s,

_ boòy ionguage. ond

3. The cont munication cycle involves encoding, transimis sion, EecocJing,

and

4. To overcome bsri iers to elf ctive communicatio ,

active _ _ _ _ _ __ is esset tial
QUESTIO
N

A Decision—management
skills

B Time management
skills

C Self- management
skills

D Self management
skills
QUESTIO
N
?'Ü P.T/ ‹I''G??S Oi SS ? i??
SICC Sü O ICC:

A Feeling
energetic

B Headache
s

C
Insomnia

o Chest
pain
QUESTIO
N

A Self—
demotivation
B Self-
awareness

C Self-
motivation

D Self-
regulation
HQ. To per form weI! ct • ork and lily in general, you must be able to nncinc ge anci
imPro /e \ ouiself in *vaiious skills. \Vl icf of the following skills helps you tc
Mriorities the tł°.ings /ou have to do to remove waste and r cłunCancy from woi k?

A. Responsibility

B. Time
management

C. Self-awareness

D. Adaptability
QUESTIO
N
C?.’: ? C Õ S C C.POS D St ? .I lC.Ü O?Ct ?.?.'OÇU.?t ?O
SS OF.‹ )? T S.Ji

A. Emotional
stress

B. Physical stress

C. Mental stress

D Social
stress
QUESTIO
N

A. Acute stress

B. Chronic
stress

C. Episodic
stress
D Physical
stress
QUESTIO
N

A
Psychological

B Physiologica
l

C Both of
them

o None of
these
QUESTIO
N

A. Physical

B. Emotiona
l

C. Mental

D. Financial
QUESTIO
N

A. Self-
awareness

B. Self-
motivation

C. Self-regulation
D Disciplin
e
QUESTIO
N

A. Disk Defragmenter

B. Restore

C Disk
Cleanup

D
Backup
QUESTIO
N

A. DrOg
onddrop

B. Double click

C. Hover

D Single
click
QUESTIO
N

A Windows
NT

B Linu
x

C Windows
2000

o Windows 3.11
QUESTIO
N

A. Windows Fax and Scan,

Settings

B. Snipping Tool, Settings

C Magnifier,
Settings

o Print, Snipping
Tool
QUESTIO
N

A Utility
Software

B Linu
x

C. Operating
system

D. BIOS
QUESTIO
N

A. Batch OS

B. Real Time
OS

C Time Sharing
OS

D Quick Sharing
OS
QUESTIO
N

A. Hover

B. Drag and drop

C Double
clicking

D Moving
QUESTIO
N

A It can Dreak

B It can stop
functioning

C It can overheat

D Data can get corrupted

QUESTIO
N

A. The processor executes more than one process at a

time

B. The progra ms a re developed by more than one person

C More than one process resides in the

memory

o A single user can execute many programs at the same

time
QUESTIO
N

A. New type file system

B. Never terminated file

system

C New technology file

system

o Non terminated file

system
QUESTIO
N

A. Only Statement
1

B. Only Statement
2
C Both Statement I and
Statement 2

o Neither Statement 1nor

Statement 2
QUESTIO
N
CI€C.OSS: .? iCJC Yi?§ O STO' F? Oi ASS6itOP.
t. GCCStGOS Ci^ '

,S€'tO (/', .S tü U S S G L (Ü)l?

O .Ut fOlSO
QUESTIO
N

A. Breaking of computer
memory

B. Allocation of resources

C Management of processes

D R e j O i n ing of broken
files
QUESTIO
N

A Detecting
viruses

B Removing \/iruses

C Detecting
worms

o None of
these
QUESTIO
N

A They clean regularly accumulating unnecessary

files.

B Cleaning files regularly helps in reduction of unnecessary space

usage.

C Helps in removing unsafe files and file

management.

o All of the
above
QUESTIO
N

A. Backup utility

B. Disk management tool

C. File management
software

D Both (B) and

(C)
QUESTIO
N

A By selling earrings to women without a

job

B By purchasing earrings from the local

market

C By buying jute from the local farmer and by providing jobs to local
women

D By attracting the women in her \/iIIage with her creative

earrings
QUESTIO
N

A Creates a new product

B Divides
income

C Manages the
business

D Takes
risks
QUESTIO
N

A. Risk—to ker

B. Low self—
confident

C. Patient

D Creativ
e
QUESTIO
N

A Independence

B Originalit
y

C
Excitement

D All of
these
QUESTIO
N
?'ï i 6ü‹.O:? üL1|: ?! O C.lT ? ??Gt' ?lOl°:? G'|
. i’ü S U ö ?

A
Organiser

B Risk—ta
ker

C. lnnovato
r

D. Motivato
r
QUESTIO
N

A. Both Statement I and Statement 2 are

correct.

B. Both Statement 1 and Statement 2 are incorrect.

C Statement 1is correct, but Statement 2 is

incorrect.

D Statement 2 is correct, but Statement 1is

incorrect.
QUESTIO
N

A. Organise
r

B. InnovCitor

C. Creator

o Producer
QUESTIO
N

A. Stimulates innovation and efficiency

B. Creates job and employment

opportunities

C Solves the problems of the

society

o Discourage the welfare of the

society
QUESTIO
N

A. Planning

B. Detailed
investigation

C. Production

D Risk-to
king
QUESTIO
N

A Makes
decisions

B Divides
income

C Takes
risk

o None of
these
QUESTIO
N

A Hardworking

B Confiden
t

C
Patient

D Trying new ideas

QUESTIO
N

A. Uncertainty of
income

B. Being Boss

C. Wealth Creation

o Freedom
QUESTIO
N

XO! ?(GB Oi OSS? LOi.(

C /
QUESTIO
N

XO! ?(GB Oi OSS? LOi.(

C /

ü ? ü ¢ C ? S ?i€l CC.
ISO?
#Q Directions n the follo ing questions, a statement of assertion (A)
s fc Io red by a stoteiaet t of recison (R) t4cii‹ the corr ct choice as.
Both cissertion (A) cmd reason (R) are true cmd ieason (P) is the correct ext
one:tion
of asscrtion (A)
(b) Both assertion (A) cmd reason (R) are tiue but recison (R) is not that core ect
expIonotion of assertion (A).
/ ssertioia (A} is true lout reason (P) is false.
Assertioin (A) is false but reoson (P) ”s true.

Asset tion (A): O'a ink o clothing business is on example of ^zage ennçIo\
ment.

Reason (P): \/\*oge emtoloyment refers to a situation where an individual

works for on empIo/er in exchange lot u iegular wa k e or soIory
QUESTIO
N

XO! ?(GB Oi OSS? LOi.(

C /
QUESTIO
N

XO! ?(GB Oi OSS? LOi.(

C /
HQ. Fill in the
Blcinks
1. Enirepi eneer cieates __ in the society.

2. An _ _ _ _ _ _ .s a |oerson who establishes a business or a venture

that Rei ero:e some value to the customer and prove to be ¡orofitaDle for
him.

3 function of an entrepren ur is directed towards tne sctisíaction

of

___ _ are dynamic ogents of change ir markets.

function of cin entrepreneur is concerned n'ith the provisions of

»’orkina Non tions anci »/elfare measures to maintoin a good »•/oik force in
tne “ organisation
A Attitud B
Risk e

c Self-
None c D
confidence
se
QUESTIO
N

A. Positive
attitude

B. Self—
confidence
C ow morale

o None of
these
QUESTIO
N

 Data Science

B. Computer Vision

C. Natural Language
Processing

D. Neural Network
QUESTIO
N

A. Computer
Vision

B. Poxel Lab

C. Pixel It

D. Pixar
QUESTIO
N

 Self-dFi\ling cars

B. Smart
Interactions

C. Face Locks

D. All of these
QUESTIO
N

A. Online
translators

B. Email filters

C Smart
Assistants

D All of
these
QUESTIO
N

Â
Twitterbot

B TwiChatBo
t

D Twitter
Robo
QUESTIO
N

 Data Privacy

B. AI access

C. AI Bias

D. Data
Exploration
QUESTIO
N

 Musical
intelligence

B. Interpersona I
intelligence

C. Linguistic intelligence

D. Mathematics I
intelligence
#Q Read the examples given belo
: Usit-›g Cha* GPT to «vrite cu
emoiI unlock techn.o!ogy of mobile phones using
Foce
camera
Turning off lights ith loT device
Hand sanitiz i dispenses having se asor
ChoOse the oot!ons thcit are not Al.

A. (i} and (ii)

B. (iii) and (i)

C. (iii) and (iv)

D. (i), (iii) and (iv)

QUESTIO
N

A Computer
Vision
B Data
Sciences

C. Natural Language Processing

D. Natural Language
Understanding
QUESTIO
N

 Data Privacy

B. Unemploymen
t

C. AI bias

D. No concern
#Q Identify the ncorrect statement(s) from the
follo» ing Deed looming is a subset of I4aclnine
Learning tv1ochine Learning is ci subset of Deep
Lecirniing Artificial Intelligence is a subset of Deep
Learning Dee/o Lee i'ning is the advanced form of
AI ancJ ML
A. Only (i)

B. (ii) and
(iii)

C. (i) and
(ii)

D. Only (iii)
QUESTIO
N

YL? tü.?:^. ClC§\ SLitS L?St iG: liis

i CCüCi

A AI Ethie
s

B Data Privacy

C Al
Bias

D AI
Access
QUESTIO
N

A. Both assertion (A) and reason (R) are true and reason (R) is the correct
explanation of assertion (A).

B. Both assertion (A) and reason (R) are true but reason (R) is not the
correct explanation of assertion (A).

C. Assertion (A) is true but reason (R} is false.

D. Assertion (A) is false but reason (R) is true.

 _ _ _ is the ability to perceive information, ancl to retain it cs kno sledge to
ñe applied to ‘lords cdoptive beño ’iour ›a ithin on en»ironrnent or context
__ _ _ _ is the in:eIIiqence where o person's ability tO r CG@l•›ise
and create sounds, rh\ thms, ord sound oatterns.
kno.^/n
cis
QUESTIO
N

l:’lu cs
i he'rsots :'ooics
 _ _ _ is o subset of .Artificial Intellig nce thot enables i»acf Ines to in-›|
orove at tasks with experience (data).
__ _ _ _ is ci domain of AI ielated to data systems utnd processes, in
I icn the system collects numerous data, maintains data sets ancJ deri 'es
meaning sense
__ _ _ _ is the domoin of AI that depicts the capnhiIit\ of o machine to
get end anal, ze isual information and ctief*NarC predict some decisions
about it.
_ _ _ _ _ _ refer to the tvDe of features e v«ont tc collec
that t.
tł cit help us in na«igatioi . are AI based
applications.
_ _ _ _ _ _ is the t\pe of into used by I^ILP
opçlications?
 _ _ _ feciture of î•JLP helps in understanding the emotions of the
șeo|oIe
mentioned »/ith the feedbac k.
HQ. Real the foIIo•.«ing tex' and ons over the io!iowinq questions based on tine
same:Reena is on encgit eering u uduate recent!y she lost her n-›obi!e thone on
th .•vcy to home. No ' she is in tne market for the n v/ one. Beinc¿ tech—
sc«vy,she's iinteresred in exolor!ng simar'phones that oiler cidvanc ü
features. Lii‹e smart locks, ' ecitïner forec astii g cnzd oti er AI
recommenclations tools.
After purchasingsmart phonc she vants to secure her moDile›/vith Al
enabled
fea:ures.
QUESTIO
N

A E\ Proble Scopin Data Data Modellin

valuation m g Exploration Acquisition g
B Problem Data Data Evaluatio Modellin
Scoping Exploration Acquisition n g

C Data Problem Data Modellin Evaluatio

Acquisition Scoping Exploration g n

D Problem Data Data Modelling —-

Scoping Acquisition Exploration E'valuation
QUESTIO
N

A. Stakeholder
s

B. Key
persons

C. Developer

D. End user
QUESTIO
N

A
Who

C. Wher
e

D. with
QUESTIO
N

A. SurUeys

B. Coding

C. Cloud
Computing

D. None of these
QUESTIO
N

A. System
Mapping

B. 4Ws Canvas

C. Data Features

D. Web Scraping
QUESTIO
N

A. Problem
Scoping

B. Data
visualization

C. Data Features

D. Web Scraping
QUESTIO
N

A UnsuperVised labelled dataset,

learning Regression

B Supervised learning — labelled data set,

Regression

C Unsupervised unla beled dataset,

learning Classification

D Supervised unlabeled data set,

learning Regression
QUESTIO
N

A ( ) is false but (R) is

true
B ( ) is true but (R) is
false

C. Both (A) and (R) are

true

D. Both () and (R) are false

QUESTIO
N

A ( ) is false but (R) is

true
B ( ) is true but (R) is
false

C. Both (A) and (R) are

true

D. Both () and (R) are false

QUESTIO
N

A Statement 1 is
True

B Statement 2 is
True

C Statement 1 and 2 both are

True

D None of these statements are

True
QUESTIO
N

A. Both Statement 1 and Statement 2 are

correct.

B. Both Statement 1 and Statement 2 are

incorrect.

C. Statement 1 is correct but Statement 2 is

incorrect

D. Statement I is incorrect but Statement 2 is

correct
QUESTIO
N

A ( ) is false but (R) is

true
B ( ) is true but (R) is
false

C. Both (A) and (R) are

true

D. Both () and (R) are false

QUESTIO
N

A Data gathered with respect to stadium, bowlers, opponent team and health is known
as Testing Data.

B Data given to anAI model to check accuracy and precision is Testing

Data.

C Training data and testing data are acquired in the Data Acquisition
stage.

D Training data is always larger as compared to testing

data.
QUESTIO
N

A Text
summarization
B Target
AdVertisements

C. Face lock in
smartphones

D. EmCliI filters
QUESTIO
N

A. Mean

B. Median

C. Mode

D. Varianc
e
QUESTIO
N

A CNN

B KN
N

C. AN
N

D. NLP
QUESTIO
N

A. NumPy

B. Pandas

C. Package
s

D. Modules
QUESTIO
N

A. Pie—chart

B. Column chart

C Bar
chart

D
histogram
QUESTIO
N

A. Median

B. Mean

C. Standard
deviation

D. Variance
İS Ci Sİm ple fİle fOrnos I that S oClta SeverateCl D'y’
StOr COmnnaS.
QUESTIO
N

D What is sum of the percentage of maths and hindi*

QUESTIO
N

A. Feature extraction

B. Object recognition

C. Image restoration

D. Image compression with data

loss
QUESTIO
N

A To enhance the aesthetic beauty of

images.
B To automate tasks involving image analysis and
interpretation.

C. To create realistic digital art..

D. To store and manage large image collections.

QUESTIO
N

A. Convolutional neura I network (CNN)

B. Recurrent neural network (RNN)

C Long short—term memory (LSTM)

network

D Hopfield
network
QUESTIO
N

A Image recognition identifies the entire image, while object detection finds
objects within the image.
specific

B Object detection requires a deeper understanding of the image content

imagetorecognition.
compared

C. Image recognition involves labelling the entire image, while object

detection assigns labels to individual objects.

D. Object detection utilizes more complex algorithms compared to image

recognition.
QUESTIO
N
G. Licco ,Jocssiç
I: .:ecücr

A. Improved accuracy in disease detection and classification.

B. Reduced workload for radiologists and other medical professionals.

C. Faster and more efficient diagnosis process.

D. All of the aDove

QUESTIO
N

A Grayscale
images

B RGB
lmages

C. Black and White

Images

D. High resolution Images

QUESTIO
N

A. Price Prediction

B. Photos albums in
smartphones

C Trend
setting

D Face lock in smartphones

HQ. Conaputei vision * forks on tne some plnenome os that __
of _ _.
#Q Peel
sonds'or
HQ. Read the follo. ing text and Ans her the io!iowinq questions based on time
some:
Sum t Sur«eillonce
A rcsidertial complexSystem
insta iled o ne z AI-keyword smcit t surveillance
system Canneros ei gcigeci ‹ith computer vision algorithms monitor common
areas ance i›sCi»idunl opcirtments upon resolve consent. The system
automuticc:fly drives susDicious activities like tiespassing kno«'ing anal
unauthorized «ehicle entry. it ti!ggersoreas cmcJ send reol-time notificcitions to
resiciants ond security personne!

Identify the computer vision applications used in smart surveillcince system

QUESTIO
N

A. 12

B. l

C. 24

D. I/l2
QUESTIO
N

A
Images

B Numerical
data

C. Graphical data

D. Text and Speech

QUESTIO
N

A. it will create real value for the organisation

B. it will create more confusion among the stakeholders

C. it is a tool that prevents the problem will not recur

D. it is a tool that prOViÖes a clear understanding of the data

flow
QUESTIO
N

A The problem can be identified and goals can be

set

B The problem con be put on paper

C The problem can be discussed among

stakeholders

D The problem can be understood by the

people
#Q *›’*hicIn ot the foilo‘/«ina statements is correct ior training
clata Dots should h relevant
Dsta should be ciutheiatic
De tc should be downloaded only

A (i) and (ii)

B
(i), (ii) and (iii)

C. (i), (iii) and

(iv)

D. All of them
HQ. *›’7hicIn ot the foiIo‘/'ina is
Testing
incorrect? (i) c'ato is the one on Inicln we irain and fit our model basicallyto fit
pcrcimeters
the
Training cioto is used only to assess performance of model
Testing dc:'o is the unse n datci for which predictions hcive to
be n c de

A (i) and (iii) only

B (i) and (ii)
only

C. (ii) and (iii)

only

D. (i), (ii) and

(iii)
QUESTIO
N

A. Problem
Scoping

B. Data
acquisition

C. Modelling

D. Searching Ideas
QUESTIO
N

A
Who

C
Where
QUESTIO
N

A. stakeholders

B. nature of the problem

C. context/situation/location

D. solution or benefits to the

stakeholders
QUESTIO
N

A
Who

C
Where
QUESTIO
N

A
Who

C
Where
QUESTIO
N

A. System mapping

B. 4Ws problem
canvas

C. Data features

D. Web Scraping
QUESTIO
N

A. designing - brainstorming — building

B. designing - deploying -
brainstorming

C. brciinstorming - designing - building

D. designing - brainstorming - building

QUESTIO
N

A Data Exploration

B E\
valuation

C. Problem
Scoping

D. Data Validation
QUESTIO
N

A Testing Data

B Training Data

C. Input data

D. None of the
above
QUESTIO
N

A. Any websites

B. Google

C. Government
Websites

D. None of these
QUESTIO
N

A Bag of
Words
B Named Entity
Recognition

C. Latent Semantic
Analysis

D. Tokenization
QUESTIO
N

A Sentiment analysis on social media

posts

B lmage recognition in self-driving

cars

C Fraud detection in financial

transactions
D Spam filtering in
email
QUESTIO
N

A. rule—based approach

B. learning- based approach

C. knowledge-based approach

D. All of these
#Q The basis of ctecision mciking depencJs
u¡oon. A zoilaLility’ of Infor
cation Pcist Exoer ience
Positi'ze Altitude
Self-Av’voieness
A. (i) and (ii)

B. (ii) and (iv)

C. (i), (ii) and (iv)

D. (i), (ii) and (iii)

QUESTIO
N

A. 4Ws Canvas

B. AI Project
Cycle

C. 17 SDGs

D. AI Models
QUESTIO
N

A. Both assertion (A) and reason (R) are true and reason (R) is the correct
explanation of assertion (A).

B. Both assertion (A) and reason (R) are true but reason (R) is not the
correct explanation of assertion (A).

C. Assertion (A) is true but reason (R} is false.

D. Assertion (A) is false but reason (R) is true.

HQ. True or False: Problem scoping in an AI urojecI only invol es iclentif›/ing
the ind!vic!uols exoeriei cing stress and understanding the natures of their
stressors, but it docs not iiz solve deteri°.Mining• here stress manifests in their
lives.

A. True

B. Fals
e
QUESTIO
N

A. Predictions

B. Confusion
Matrix

C Realit
y

D
Recall
QUESTIO
N

A True Negative, False

Negative

B True Negative, False

Positive

C True Positive, True

Negative

D True Positive, False

Positive
QUESTIO
N

A. Precision

B. Predictio
n

C. Precision

D. Recall
QUESTIO
N

A
Precision

B Accurac
y

C. Predictio
n

D. Recall
QUESTIO
N

A
Prediction

B Accurac
y

C. Reality

D. F1 Score
QUESTIO
N

A. defined as the fraction of positiVe cases that are correc tly identified.

B. defined as the percentage of true positive cases versus all the cases
where the prediction is true.

C. defined as the percentage of correct predictions out of all the

observations.

D. comparison between the prediction and reality

QUESTIO
N

A. Accuracy and
Precision

B. Precision and Recall

C. Recall and Accuracy

D. Recall and Reality

QUESTIO
N

A viral
outbreak
B forest
fire

C fIOO
Ö

D spam
filter
QUESTIO
N

A False
Positive

B True
Positive

C. False Negative

D. True
Negative
QUESTIO
N

A. Precision and
Accuracy

B. Precision and Recall

C Accuracy and Recall

D Precision, FI
score
Topic:
Summary
Topic: Homework From ACE
Module
SCORE :
95%
 Artificial_Intelligence (AI) is the ability o.f..machines o+r. computer -sys-tems to perfor—m tasks that typically
intelligence. suc as earning. proAem-solving. decision-making. and understanding language
require human

Imagine a voice assistant like Alexa or Google Assistant. When you ask. "What is the weather today*” it
understands
your question searches for the latest weather report. and gives you an accurate answer.

1 Learning from Data (Machine Learning) — AI learns from a vast amount of past data For example.
a virtual assistant improves its responses by learning from millions of conversations
2. Recognizing Patterns — If you ask for the weather every morning, the assistant can predict Ihat you
mighl ask
the same question at the same time daily and even remind you
3 Decision Making — Based on the question. it decides the best response to provide. just like a human
would
HQ. Ak I zt ts to learn ìñO N e the pi oblenn for on Al Lh
to s Projec e
The 4W Problem Canvas helps in the ke_eIemenIs of a problem by answering four
identifyin questions:

Example: Sludenls struggling to understand mathema¿i cepts

2. ‘‹’› !!r.—t eat is the exact problem? (The challenge or
ssue)
. F ampleStudents find it difficult to uMerstand complex math problems and need
personalized solutions
3
4. •':\r i } Where does the problem occurs (Location or scenario)
Example: In online learning platforms or in schools with large class sizes
Why is this problem important?
future (Impact and need for a solution)
Example. If students do not understand math concepts. their academic performance may suPer.
opportunities.
affecting
A well-structured problem statement helps in defining the goal of the AI project A common template is:

“The problem of [describe the problem] affects [who is affected], impacting (why it is important]. A
successful solution would [describe the ideal Al-based solution]."

"The problem of students struggling with math problem-solving aWects school students. impacting their learning
outcomes and confidence A successful solution would be an Al-based chatbol that provides step-by-step
explanations for math problems.“
AI learns from data using different approaches The given images illustrate two key learning
methods:

• In Supervised Learning. AI is trained using labeled data. where each input has a corresponding
output
• Figure 1 (Left Side) shows images of dogs and cats with labels. helping AI learn to classify
them
correctly.
• Example: A model trained on labeled dog and cat images can later identify new animal images
accurately.
• In Unsupervised Learning. AI is gi\/en unlabeled data and must identify patterns on its own
• Figure 1 (Right Side) shows an AI system receiving an unknown input and predicting it belongs
to the "dog" category based on similarities.
• Figure 2 further explains this concept—AI does not know the names of the objects but can
differentiate
groups ‹dogs vs. cats) based on patterns
• Example: AI clustering customers into groups based on shopping behavior without predefined
labels
Computers process information in binary. whereas human languages are complex and full of variations.
Teaching to understand and interact in human languages presents several
computers
challenges
1. .4 ›:l:i::u !\’ 1 F.'ri!*i¡ I' \' .‹i !i›\ s Words often have different meanings based on
context. Example: "Bank" can mean a financial institution or the side of a river.

*"
2 =n°a”.*r’ñ ?ti ‹ictt!rnj . .
ir * re¿zes Different languages follow different grammar
rules. making it hard for AI to profss them accurately. Example: "Red car" (English) vs. "\/oiture rouge"
(French)

4. °=1›: fiA cfs Sir. Language differs by region and culture. Example. "Lift" (UK) vs.
"Elevator" (US). Al requires large datasets to recognize variations
HQ. Ån Ål cicłe time following scles medic tion tor a new' mobile
model oîaon , which
!iii Üalculating Precision. Rec ali, a ncI F1 č core

• Precision = TP TP + FP)
= 50 I (50 *“40) = 50 1 90 = 0.5556

• Recall = TP / (TP + FN)

= 50 I (50 + 12) = 50 162 = 0.8065

• F1 Score = 2 • (Precision • Recall) / (Precision + Recall)

=2 (0.5556 • 0.8065) / (0.5556 *0.8065)
= 2 • (0.4481) / 1.3621
= 0.6572 (approx)
HQ. You tell youi eo stre d
younger s otn AI. Cen ing o A sugges s
you exileii-‹ eos ,
AI-powered recommendation systems analyze user behavior to suggest videos. This process involves

1 ' °: c Data oiled.*i ¿q/ AI tracks watch history. likes search queries. and time spent on videos
2. obh‹l.mi,-‹t—— - ii , It compares users with similar preferences and suggests videos liked by
others with similar tastes
3. ^/r+°,'c - t-C.ase I '.I'° ning— Al analyzes video content
(titles. descriptions genres) and recommends similar vef o b s
4. ^cxc Lead ni «i Phi eon liz‹Itio's Advanced AI models like neural networks detect patterns.
preferences. and continuously refine recommendations
predict
HQ. You are tasked « ith building c1n AI model to pt edict house prices bcised
on Iocc:tion, size and other features.

(A) Eoev «i I you structure the problem using the 4W Problem

Canvas?

(s) \^/hut kincl of data v' ill you need to acquire for the solution?
1 Who — Homebuyers. real estate agents. and property investors who need accurate price
predictions
2. What — An AI model that predicts house prices based on factors like location. size, and features
3. Where — The real estate industry. including online property platforms and valuation agencies
4 Why — Helps buyers make informed decisions. sellers set competitive prices. and investors analyze
market
trends
(E I Data Re ciuire cl for the SoIutioiz

1 Property Features — Location. size (sq ft ). number of bedrooms. bathrooms. and amenities
2 Market Data — Historical prices, demand trends. and economic factors
3. Neighborhood Information — Crime rates. school ratings. transportation facilities
4. Real Estate Listings — Current asking prices and time on the market
 st :: . c‘ represents Supervised Learning. where the model is trained using labeled data. The system
learns from examples where both input (image of a dog) and output (label "DOG") are provided allowing it
to classify new inputs correctly.

! “*¿ B? co' cJ I:?,I ! represents

Unsupervised Learning. where the model is not given labeled data Instead. it
identifies
patterns and differences among inputs grouping similar items (dogs and cats) without predefined
categories
HQ. Meenu uses a lanquacge learning ago powered loy’ AI. She tyoes the scntence
"I on point to the hoink" and the atop coirects it by suggesting "I ann coi‹ng
to the ri'/er bork" ins:ead oi 'bank' as a financial institution. Why mighi this
misunderstanding happen in on AI model?
This misunderstanding happens due to word ambiguity in Natural
Language
The Processing
Al model relies on context (NLP)
to determine meaning. but if it lacks
enough
understanding.contextual
it may assign an incorrect interpretation.
meanings— a financial institution
In this case and a riverbank
"bank" has The AI might have been trained on a dataset
multiple
where "bank” was more commonly associated with rivers. leading to the
wrong suggestion
Improving contextual
architectures can help analysis with larger
reduce such errors datasets and advanced
models like transformer-based
HQ. / healthcare s/stent predicts if a patient hers a certain
The confusion matrix is given
disease
below:
Real“ ..Disease Real" No Disease
Prediction: Disease 40 25
Prediction: No Diseoše ‹J5; 120

(i) Ccilculate the systiv's lot se I egotive iate

(FNR).

(ii */‹*no t is the precision of the systerri?

FNR is calculated
as

(ii) Precision

Precision is calculated as:

Thus.
• False Negative Rate (FNR) =
27.27%
• Precision = 61.54%
Machine Learning is a branch ol Anificial Intelligence that enables computers to learn from data and
improve their performance without explicit programming

How a Chatbot Learns to Respond Intelligently:

1. 3uta T•a”n.t. : The chatbot is trained on large datasets containing various conversations and
labeled responses
2 Nature I Lo: .age roe ssi!‹¿ \.hLP}: It analyzes user input. understands intent. and extracts key
information
3. F,1r -.ol Lr n.! sin‹ : ML models like deep learning and reinforcement learning help the chatbot generate
accurate responses.
4 cu:-ir uc›u In°t i ‹a .'e.'\\ cilt: The chatbot refines its answers based on user feedback and repeated
interactions
HQ. Your t e r n «cintsto create a voice assistant thot heltos visua fly inn|ca irecl
usut s

(A) Explciin Ino•a‹ the |orojccI •NOC\D fit into the Al Project Cyc e.

(b) \^/hut challenges could orise during problem sco¡oinq for this
project?
1 Problem Scoping: Define the problem. identify user needs and set objectives for the voice assistant.
Ensure it
helps visually impaired users navigate apps effectively.
2 Data Acquisition: Collect voice commands. user navigation data. and app structures to train the model
3. Data Exploration & Preprocessing: Clean and analyze the data to recognize speech patterns and app
interactions
4. Model Building: Develop and train AI models using NLP and Speech Recognition for accurate voice
assistance
5. Evaluation & Deployment: Test accuracy. improve based on user feedback. and integrate into mobile
applications
1 User Diversity: Different accents. speech patterns. and command styles must be understood
2. Technical Constraints: Handling background noise, speech misinterpretation. and app compatibility
issues
3. Privacy Concerns: Ensuring data security while collecting and processing user voice inputs
4 Navigation Complex ity: Adapting to various mobile apps with different layouts and functions
The two figures represent Supervised Learning and Unsupervised Learning in Machine Learning

1 * \incmiscrf Learnirirt îTo t› F it \irc)

The model is trained using labeled data. where input-output pairs are known
The system learns from examples (e g . rabbits labeled correctly) and applies this learning to
new inputs
This approach is used in applications like image classification and spam detection

The model is not given labeled data but identifies patterns and groups similar objects.
Here, the system clusters difterent animals based on similarities. even without knowing their
labels
This method is used in customer segmentation and anomaly detection
 HQ. Sore talks to leer phone assistant,asking it to olay hei favorite song.
Ho•vever,
\^/hat couldofhave
instead gonewrong
alas ink it!a on
' The \A*heels thethe
Al assistant's understandinc,
Bus”, it oloys and ho»/
sotcethinct coivpleteI\
do different.
such syStClS *NOr k \o und rstancl human language?
The Al assistant may have misunderstood Sara s request due to Automatic Speech Recognition (ASR)
errors, language processing limitations, or ambiguous pronunciation

Speech Recognition Errors. Misinterpretation due to unclear pronunciation or background

noise Homophones & Ambiguity: Words sounding similar may lead to incorrect song
selection
Limited Context Understanding. The AI might not recognize Sara s favorite song if her
preference data is
incomplete

ASR (Automatic Speech Recognition) converts speech into text.

NLP (Natural Language Processing) interprets the meaning of words and context
ML & AI Models use previous interactions and user preferences to refine responses
Error Handling I¥lechanisms Some assistants ask for clarification if uncertainty is detected
HQ. In self-clriving cors, AI oIays o significant role in ana I\ zinc real-time üotu tronn
Lhe
envii oi-›v ent. *.Vhat is the primary AI technique used to make dri«in¿ decision
s?
The primary Al technique used in self-driving cars to make driving decisions is Deep Learning with
Reinforcement Learning

Uses Convolutional Neural Networks (CNNs) to process images from cameras. detecting objects
like
pedestrians. traffic signs. and lanes
Recurrent Neural Networks (RNNs) help in predicting traffic patterns and movements

The car learns by trial and error through simulation and real-world driving
Uses reward-based learning to optimize decisions like braking. acceleration. and lane changes.
3 S ens or F as ion d‹ Dec ision ffi a kin g
Combines data from LiDAR, cameras, GPS, and radar to create a real-time driving
model AI applies path planning algorithms to navigate safely while following traffic
rules
Ramesh should apply Unsupervised Learning. specifically Clustering Algorithms to group similar
customers
together.

1 V/!:y Ur›su; er /is c cl LearMinq?

The dataset does not have labels (customer types are unknown)
The goal is to find hidden patterns in the data

K-Means Clustering: Divides customers into K groups based on similarities

Hierarchical Clusterin g Creates a tree-like structure for grouping
customers
DBSCAN: Detects customer groups based on density without needing d
fixed number of clusters.
Helps in targeted marketing by identifying customer preferences
Assists in personalized recommendations and business decision-
making
HQ. Piiyu is integraling datci front mLJltigle sources into iner Al system. !
Nha\ cM e t‘zvo ohciIIe:ages she might face ciut ing date integration?
Priya might face the following two challenges during data integration in her Al
system

Data from different sources may be in varied formats (e.g. CSV JSON, SQL databases)
Differences in date formats, numerical precision, or missing values can create integration
issues

Some records may be repeated across sources, leading to biased AI predictions

Duplicate data needs to be identified and removed for accurate model training
A face lock system in a smartphone works using Al-based facial recognition technology that follows
these steps.

The smartphone's front camera captures the user's

face.
AI models detect key facial features like eyes. nose. and jawline

The system converts the face into a mathematical representation using techniques like neural
networks or
computer vision
Unique patterns. such as distance between eyes or facial contours are extracted

The extracted features are compared with the stored facial data using an AI model
If the similarity exceeds a threshold. access is granted

Infrared sensors or depth detection prevent unlocking using photos or videos

Some systems use 3D facial mapping for added security.
Converting text into a common case (lowercase or uppercase) is significant in Natural Language
Processing
(NLP) and text processing because

Words like "AI" and "ai" are treated as the same. reducing data variability

Case differences can cause mismatches in search engines, chatbots, and text analysis

Case normalization simplifies text comparison and tokenization making algorithms more
effi
cient
^ Prevcrts Rerlrmúan! Entries
Helps in sentiment analysis, spam filtering, and keyword extraction by treating words uniformly.
A confusion matrix is a table used to evaluate the performance of a classification model. Given the
values:
• True Positives (TP) = 150
• False Positives (FP) = 30
• True Negatives (TN) = 100
• False Negatives (FN) = 20

The confusion matrix is represented as:

HQ. Your cousin uses an ml-povverecTrobot vacuum cleaner. Cen /ou exolciin
I om' this robot Iecrns to elect. rooms efficiently using AIR
An AI-powered robot vacuum cleaner learns to clean rooms efficiently using Machine Learning (ML)
and Sensor-based Navigation

Uses Simultaneous Localization and Mapping (SLAM) to create a virtual map of the room
Sensors like LiDAR and cameras detect furniture. walls. and obstacles
2 Path OptinJizaticn
AI analyzes past cleaning pallerns and optimizes routes for better
coverage Avoids repeated cleaning of the same spot and prioritizes
dusty areas
3 Ohs•ac le Detc°ction & A’.oicla nce
Uses computer vision and infrared sensors to detect objects and avoid
collisions Loai ninII Uscr Prefer nccs
AI remembers high-traffic areas and adjusts cleaning
frequency. Can integrate with voice assistants for scheduled
cleaning
Zara can use the 4W Problem Canvas to define her Al-based student engagement tracking
app:

Students who struggle with engagement

Teachers who find it difficult to monitor student attentiveness

Students get distracted. reducing learning effectiveness.

Teachers lack real-tame insights into student engagement

In classrooms (online/offline) where students attend

lectures During self-study sessions when students lose
focus

Boosts learning outcomes by identifying engagement

levels
Helps teachers adapt teaching methods for better student
involvement
The diagram illustrates the key components of an Al system. which include:

1 T aii° ii c: f.’r ‹ir. AI models are trained to estimate largel values using historical data
2. E'v.‹ Iurl i i: .›I 'J J ie The accuracy of trained models is assessed before deployment
3. E stin:\ie : F on icns The system estimates key parameters like output power, coupling
coefficient, and load resistance for decision-making
4 D c c.si a-'Za•:ird Pr cc .s Based on the estimates. the system determines
whether to turn on or off an external process. ensuring optimal operation.
HQ. Rasoul‘s srananaother loves cooking and writes recipes in ther native langciace.
e/onts to de. eloa on AI s›/stew tl ic:t con»erts her handv^ ritten recides
Ranc into
I English.
\^/!aot
translcite it intochaI!enqes might the
another langucge? AI face Eileen trying io
reoc! hands iiten text oizd
Rahul s AI system faces several challenges in reading and translating handwritten recipes

1 I aisri .’i"*ir ‹j ‘lbi I:il!'\' Different handwriting styles, slants. and inconsistencies make recognition
difficult
2. ñ oin, ‹:i C.r!r rl “F-xt Smudges faded ink. and unclear characters can reduce accuracy.
3. C i : x*uat Und‹: ! si s::¢!i'› AI must correctly interpret abbreviations. symbols. and regional
phrases before
translation
4 Trei-s Intior .«c°.ui ac ›' Translating from a native language to English requires maintaining grammatical
correctness and meaning while handling multiple word meanings
HQ. A machine lear wing wnodeI is de .•elooed ’Inetiner students will ness or
to medic t on exa r n. The confusion matrix is fail
Reoli . 'Real :
Predictiori : Pass “ 3 Fail
Pass 70 0
Prediction : No 20 8
(i) hentify
Foil the numloer of true posit v s (TP) and false
0 positives
(F T )
(ii Calcu ate time precision cmd recall foi the
) model
(i) True Positives (TP) & False Positives (FP):
• TP (Correct Pass Predictions): 70
• FP (Wrong Pass Predictions): 30
(ii)Precision & Recall Calculation:
• Precision = TP/TP+FP = 70/100=0.7 (70%)
Out of all "Pass" predictions. 70% were correct

• Recall = TP/TP+ FN=70/90=0 778(77.8%)

— Out of all actual "Pass" students. the model correctly predicted
77.8%
1 Who: Homebuyers. sellers.real estate agents and investors
2 What: Predict house prices useng AI basea on location. size and features
3 Where: Real estate industry. online property plattorms. urban and suDurban
areas
4 Why: Helps in better decision-making. accurate pricing. and market analysis

1 Property Features: Location. size. number of rooms amenities

2 Market Data: Recent sales demand-supply trends
economic indicators
3 Historical Data: Past house prices. inflation adjusted rates
4 External Factors: Interest rates. crime rates. infrastructure
projects
The image shows a robotic arm playing chess. which represents Artificial Narrow Intelligence (ANI) or
Weak Al

• This AI specializes in a single task (playing chess)

• It follows pre-defined rules and learns patterns to improve performance
• Unlike humans. it lacks general intelligence and cannot perform tasks beyond chess
• Famous examples include IBM's Deep Blue and AlphaZero. which use AI algorithms to play and win
chess games
HQ. Rim is ci student vvtno loves interacting with cJifferent oeoo!e arounci the ' 'or!d
She uses o translation off to chot n ith her friends front Jopan and BroziI. Ho
ve Rei, she noticed that sometimes the translations don't capture the real
meaninc oi her
^/ords. \\*Iay do you think this hapoens? HORN does AI struggle to unct<°rstand context
AI struggles to understand context in human languages due to the following challenges:

1 Ambiguity in Language: Words can have multiple meanings depending on the context. For example,
"bank" can
mean a financial institution or the side of a river. Al may misinterpret such words.

2 Cultural and Contextual Differences: Some phrases or idioms do not have direct translations. For instance. an
English phrase like "break a leg” means good luck. but a direct translation may confuse non-English spea
kers

3. G ramma r and Sentence Structure Variations: Different languages follow different grammatical
rules
AI sometimes struggles to maintain the correct meaning when translating between languages with vastly
different structures. Iike Japanese and English

4. Lack of Emotional and Tone Detection: AI may not recognize sarcasm humor. or emotions
correctly. For example. “Oh. great!"
could be genuine praise or sarcasm. but Al might not differentiate
HQ. Your friend receivecł a personalised slaopÇinq
recommendation fromon online stoi e. Ho» coes AI help in giving such
iecommericlations?
AI helps in giving personalized shopping recommendations through the following methods:
1 Analyzing User Behavior: Al tracks browsing history. past purchases. and time spent on different products
to
understand user preferences

2 Collaborative Filtering: It compares a user's activity with similar users to suggest products that others with
similar interests have purchased

3. Content-Based F iltering: Al examines product details (brand. categoryprice. etc ) and recommends similar
products based on past purchases

4 Real-Time Data Processing: AI processes user interactions in real time updating recommendations
dynamically
based on recent searches or clicks

5 Use of Machine Learning and Deep Learning: Advanced Al models analyze large datasets to predict what
a
user might Iike based on patterns and trends
QUESTIO
N

(B) ür!¿jos' e::!ors :o co.!cc‹ rein c Co'c lGr L:is

o o;ec'
1 Unclear Objectives: Without a well-defined goal the data collected may not align with the intended sentiment
categories
positive‹e g negative neutral)
2 Biased or Incomplete Data: Poor scoping may lead to collecting data from limited sources. missing diverse
linguistic expressions slang or regional variations
3 Overfitting or Underfitting: If lhe scope is loo narrow. trje model may fail on real-world data. if too broad. it
may introduce
unnecessary noise
4 Privacy and Ethical Issues: \Mtnout proper scoping. sensitive or personal data might be collected wtnout user consent.
leading
to compiıance issues

1 Social Media APis: Use platforms like Twitter. Facebook or Instagram to collect real-time user posts and comments
2 Web Scraping: Extract public user reviews. blog comments and discussion forums to gather diverse sentiment-based
text
3 Surveys and Feedback Forms: Collect direct user input with labeled sentiment responses for better accuracy
4 Existing Sentiment Datasets: Use pre-labeled datasets like IMDb reviews Amazon product reviews or Sentiment
140 for model
training
5 Crowdsourcing: Leverage human annotators to label data ensuring high-quality and contextually accurate sentiment
tagging
Answers

Defini

Exam
ț

Hum
a
Invol
v

Appli
c
Limited Data AI may next have aCcurate or updated infornJatic+' about paiks Cth
Understanding
Keyword Matching Instead lakes
AI might fc'cus on "closest park” and ignore "v•ath a
of lake "

Lack of Real World Ai may net cc'mprehend user intent Iike a human
Knowledge won Id
Training Data If AI was trained on incc›mpIete or biased õata it may gave
Bias incorrect
HQ. Due iug o iaimily dimmer, your uncle mentions that self-driving cars cire go
tered lay AI. NON do these cors use Al to drive safely?
Self-driving cars rely on AI to analyze their surroundings and make real-time
decisions. The key AI components thot help ensure safe driving
driving
include:

Compwtor vi
sion

Leaming
Tara s problem statement for predicting food shortages in rural areas should include the following elements:

1 Problem Definition: Clearly state the challenge of food shortages in rural areas and the need for early
prediction
2. Data Requirements: Identify key data sources such as weather patterns. crop yields. soil conditions. and
historical food shortage records.
3. Al Approach: Mention the use of machine learning. satellite imaging. and predictive analytics to analyze
trends
and forecast shortages.
4. Impact & Goals: Highlight how Al predactions can ensure food security. enable timely aid distribution, and
suppon
rural sustainability.
Role of Neural Networks in AI

1 Portein Fa yjnit.o Neural networks help in recognizing patterns from large datasets.such
as facial recognition in security systems
2 ñ ec si oiJ-'1z:.Iii‹ They process complex data inputs and make intelligent decisions. like
detecting fraud in
banking transactions
3 Lea rnin frsr. Da tn Neural networks improve their accuracy over time through training. making them
useful for applications like self-driving cars
4 E xa!•1|;ie: Ii ac,o la s sifi z» t:onA neural network can classify images as cats or dogs by learning
from
thousands of labeled images
1. I :‹*ñ \*inric.I . ,us:'tr‹i tc. Used in Al based chatbots like Siri. Alexa. and Google Assistant for human-
otusinteractionslike
2. I.1z‹ . i':. I! .a: ! •i Helps in translating languages automatically. such as Google Translate
3 Sentim'. t ›1na i,’sis Analyzes text to determine emotions. used in customer feedback and social
monitoring
media
4 S ,oe eo!: Re cr›r n‹t:c.n Converts spoken language into text. used in voice assistants and automated
transcription
services
1 Peech Recognition — Converts spoken words into text using NLP (Natural Language
2 Natural Language Understanding (NLU) — Analyzes the text to identify intent and meaning
Processing)
3 Machine Learning & AI Models — Uses vast datasets to improve accuracy and personalize responses
4 Response Generation — Processes the request and delivers appropriatereplies or actions. such as
alarms or playing music.
setting
HQ. Rinci is working on an AI project to imorove traffic management in nei cit\•.
v.'ìti
Help hcr the foIio'a ing.

(a) Expiciin the importance of identifying stckeholcJers

(b) I-ion can defining success criteria hells in Droblem

scoping?
1 Understanding Needs — Helps address the concerns of authorities commuters and
emergency services
2. Resource Allocation — Ensures efficient use of budget. technology. and manpower.
3. Collaboration & Support — Gains approval from government bodies and traffic departments
4 Sustainable Implementation — Ensures long-term success by considering public and environmental
factors.

1 Clear Objectives — Establishes measurable goals like reduced congestion or improved traffic
flow
2 Performance Evaluation — Helps assess Ihe AI model's effectiveness through KPls
3 Efficient Decision-Making — Guides the selection of the right Al techniques and data sources
4. Stakeholder Satisfaction — Ensures the solution meets expectations of authorities and the public.
Deep learning is a subset of machine learning that uses artificial neural networks (ANNs) to mimic the way
humans
learn from data. The given visual representation illustrates a multi-layered neural network. which is
deep learning
fundamental to
1 in izr:\ La , This layer receives raw data (features like X1. X2) and passes it to the next layer
Ü—! I i':I”. La, mrs The network contains multiple hidden layers where neurons process the input
2
by applying weights.biases. and activation functions More layers enable the model to learn
complex patterns
3 Bi:‹s Rei. r›:‹s These help in improving learning by allowing the model to shift the activation
function as needed
4 C t zu* La\ er The final processed information is passed to the output neuron. which gives the
final prediction or
classification
HQ. / ryan 'oice-activated Al to search for intormation online. He asked, '?
usescthe toIlest Lui!dit g in Kolkata'? ancJ got an cinswer cboutc Luilding in
*hat‘s
crotlser city.
\^/lay might the A! have made this mistake, ance ho» doesAI process
queries to oio'zide relevcint information?
1. t:ice:! F AI may misinterpret spoken words due to accents. background noise. or
cc›.jriiiici Si rem unclear
pronunciation
2 .\ Is ‹: t;’ inu'<ip Pi The AI might not correctly recognize "Kolkata" and instead prioritize
o.Yessin popular results globally

If the AI lacks geolocation awareness. it may fetch information

3 ¡_@¿¿ti ,{q ¿',[,. ¿*
from
*,1{¿¡nt°-rtar ta-: i another city
4. rather
r:tc!a-c ‹I or
than Ii ›a::r iir::to DaI The Al may retrieve incorrect or outdated information from its dalabase or
Kolkata
the
internet
5 !c',’ F recess!no, S t‹ tas AI converts speech to text. extracts keywords. identifies intent. searches
databases ranks results. and provides the best possible answer.
Topic: Homework From ACE
Module