ARTIFICIAL INTELLIGENCE

BANGALORE UNIVERSITY – SEPTEMBER 2023 (SOLVED QP)

SECTION - A

Answer any 5 questions. Each carries 2 marks. (5 x 2 = 10)

1) Why visualizing the model architecture is important ?

Ans. Here are two key reasons why visualizing model architecture in AI is important:

1. Understanding the Model: Complex AI models, especially deep neural networks,

can be like black boxes. Visualization helps you see the data flow, layers, and
connections within the model. This transparency allows you to understand how the
model processes information and arrives at its outputs.
2. Improving Model Performance: Visualization can reveal inefficiencies in your
model architecture. By seeing where bottlenecks or redundant computations occur,
you can optimize the structure. This can lead to improvements in accuracy, efficiency,
and overall model performance.

2) Define Artificial Intelligence

Ans. Artificial Intelligence refers to systems or machines that mimic human intelligence to
perform tasks and can improve themselves based on the information they collect.

AI is also the field of study that develops these intelligent machines. This field explores
techniques like machine learning and neural networks to achieve machine intelligence.

3) Which is same as human vision? Describe.

Ans. In AI, the concept most similar to human vision is Computer Vision. Here’s why:

1. Image and Video Analysis: Computer Vision enables machines to interpret and
process visual information from the world in a manner similar to how humans use
their vision. This includes tasks like object detection, image classification and facial
recognition, allowing AI systems to analyze and understand visual data.
2. Mimicking Human Perception: Through techniques such as convolutional neural
networks (CNNs), Computer Vision models can mimic the way human brains process
visual information, identifying patterns, shapes, and structures within images. This
capability is crucial for applications such as autonomous driving, medical image
analysis, and augmented reality, where understanding and interacting with visual
inputs is essential.

4) What is an Optical Character Recognition (OCR)?

Ans. Optical Character Recognition (OCR) is a technology used to convert different types
of documents, such as scanned paper documents, PDF files, or images captured by a digital
camera, into editable and searchable data. OCR involves recognizing text characters in an
image and converting them into machine-readable text.

Key Aspects of OCR:

1. Text Recognition: OCR systems analyze the shapes of characters and words in an
image, recognizing patterns that correspond to specific letters and numbers. This
allows for the extraction of text from images, making it possible to edit, search, and
manipulate the text as if it were typed on a computer.
 2. Applications: OCR is widely used in various applications, such as digitizing printed
documents, automating data entry, enabling text-to-speech conversion for visually
impaired individuals, and improving searchability in digital libraries and archives. It
plays a crucial role in document management, data extraction, and streamlining
business processes by reducing the need for manual data entry

5) Why Power - Bl is using instead of Excel ?

Ans. Microsoft Power BI is often chosen over Excel for certain tasks due to its advanced
capabilities and features that are specifically designed for data analysis and business
intelligence. Here are two key reasons why Power BI is preferred over Excel:

1. Advanced Data Visualization: Power BI provides advanced data visualization tools

that allow users to create interactive and dynamic dashboards. Users can drill down
into data, filter information on the fly, and gain deeper insights through visual
exploration.
2. Data Integration: Power BI can connect to a variety of data sources including
databases, cloud services, and APIs, allowing for seamless data integration and
comprehensive data analysis across different platforms.
3. Real-Time Data and Large Data Sets: Power BI is built to handle large volumes of
data and provides real-time data processing capabilities. This scalability makes it
suitable for enterprise-level data analysis and reporting, where Excel might struggle
with performance issues when dealing with very large datasets.

6) Give any four options are available under the format tab. (In Power BI)

Ans. In Power BI, the Format tab provides various options to customize the appearance and
functionality of visualizations. Here are four options commonly available under the Format
tab in Power BI:

1. Title:
oAllows you to add, edit, and format the title of the visualization.
oOptions include font size, color, background color, alignment, and text style.
2. Data Labels:
o Enables the display of data values on charts and graphs.
o Options include label position, color, font size, and formatting options.
3. Colours:
o Gives control over the colors used throughout your visualization.
o Define colors for data series, backgrounds, borders, and text.
4. Axes:
o Provides settings for the x-axis and y-axis of charts.
o Options include axis titles, scale (logarithmic or linear), start and end values,
font size, color, and gridlines.

7) What are the various line types in Data Visualisation?

Ans. In data visualization, various line types are used to represent different aspects of data
and to enhance the readability and interpretability of charts. Here are some common line
types:

1. Solid Line:
o A continuous, unbroken line.
o Often used to represent a direct trend or continuous data in line charts.
2. Dashed Line:
o A line made up of dashes with spaces in between.
o Typically used to represent projections, forecasts, or secondary data series.
o
 3. Dotted Line:
o A line consisting of small, evenly spaced dots.
o Used for representing less certain data, secondary trends or comparative
baselines.
4. Dash-Dot Line:
o A line with alternating dashes and dots.
o Used for emphasizing specific data series or distinguishing between multiple
data sets in the same chart.
5. Double Line:
o Two parallel lines.
o Used less frequently, but can represent boundaries, thresholds or highlighted
ranges.

8) What is the difference between Categorical Analysis type and Continuous Analysis type ?

Ans: The points of differences between are as follows:

Point of Categorical Analysis type Continuous Analysis type

Difference
1. Type of Represents data that falls into Represents numerical data with
Data distinct, non-numerical categories. infinite values between any two
Examples include customer region points. Examples include sales
(e.g., North, South, East, West), figures, temperature readings, or
product type (e.g., Shirt, Pants, time measurements.
Shoes), or yes/no answers.
2. Analysis Focuses on identifying patterns, Focuses on relationships and
Focus trends, and comparisons between trends within the numerical data
the different categories. This often itself. This might involve
involves techniques like calculating measures like averages,
frequencies, percentages, or using minimum/maximum values, or
visualizations like bar charts or pie using visualizations like line
charts. charts or scatter plots.

SECTION - B

Answer any 4 questions. Each carries 5 marks. (4 x 5 = 20)

9) Explain the key workloads of Artificial Intelligence.

Ans: Artificial intelligence (AI) tackles a wide range of tasks, but some key workloads
emerge as foundational:

1. Machine Learning (ML): This is the core engine of many AI applications. ML

algorithms learn from data, allowing them to improve performance on a specific task
without explicit programming. Tasks like image recognition, spam filtering, and
product recommendations all rely on machine learning.
2. Deep Learning (DL): A subfield of ML, deep learning uses artificial neural networks
inspired by the human brain. These complex networks can learn intricate patterns
from massive datasets, enabling applications like speech recognition, natural language
processing, and self-driving cars.
3. Natural Language Processing (NLP): This field focuses on enabling machines to
understand and manipulate human language. NLP tasks include sentiment analysis,
machine translation, chatbot development, and summarizing text documents.
4. Computer Vision (CV): As discussed earlier, computer vision equips machines with
the ability to interpret visual information. CV applications include object detection in
images and videos, facial recognition, medical image analysis, and autonomous robot
navigation.
 5. Robotics : AI plays a vital role in modern robotics. By combining computer vision,
sensor data processing, and decision-making algorithms, robots can navigate complex
environments, perform intricate tasks, and interact with the physical world. This is
used in industrial automation, surgical robotics, and even personal assistants like
vacuuming robots.
6. Anomaly Detection: AI excels at identifying patterns in data. Anomaly detection
uses these capabilities to find deviations from normal patterns, which can be helpful
in fraud detection, system failure prediction, and identifying unusual network activity
for cybersecurity purposes.
7. Generative AI: This emerging field focuses on AI models that can generate new
content, like images, text, or music. This has applications in creative content
development, product design, and even drug discovery through molecule generation.

10) Explore Computer Vision with Examples.

Ans: Computer vision is a field of artificial intelligence (AI) that enables computers and
systems to derive meaningful information from digital images, videos, and other visual
inputs. Here are some key areas of computer vision with examples:

1. Image Classification

Example: Classifying images of animals into categories such as dogs, cats, and birds.

Application: Automated tagging of images in social media platforms, organizing photo

libraries.

2. Object Detection

Example: Identifying and locating objects within an image, such as detecting cars,
pedestrians, and traffic signs in a street scene.

Application: Autonomous vehicles use object detection to navigate and avoid obstacles.

3. Image Segmentation

Example: Dividing an image into segments to isolate regions of interest, such as segmenting
medical images to identify tumors.

Application: Medical imaging, where precise segmentation is crucial for diagnosis and
treatment planning.

4. Facial Recognition

Example: Identifying or verifying individuals based on their facial features.

Application: Security systems for access control, user authentication in smartphones.

5. Optical Character Recognition (OCR)

Example: Extracting text from scanned documents or images of text, such as reading the text
from a photographed page of a book.

Application: Digitizing printed documents, enabling searchability and text editing.

6. Image Restoration

Example: Enhancing the quality of images, such as removing noise from a photograph or
restoring old, damaged photos.

Application: Improving the quality of low-resolution images, preserving historical

photographs.

7. Pose Estimation

Example: Estimating the pose of a person in an image, such as determining the position and
orientation of limbs in a sports video.

Application: Motion capture for animation and gaming, analyzing athletic performance.

8. Scene Understanding

Example: Analyzing the entire scene to understand the relationships between objects, such as
identifying a living room and recognizing the furniture and people in it.

Application: Robotics, where robots need to understand their environment to interact with
it effectively.

9. Augmented Reality (AR)

Example: Overlaying digital information on the real world, such as displaying navigation
directions on a live view of the street.

Application: AR apps like Pokémon GO, AR navigation in smartphones.

10. Medical Imaging

Example: Analyzing X-rays, MRI scans, or CT scans to detect abnormalities, such as

identifying tumors or fractures.

Application: Assisting radiologists in diagnosing diseases and planning treatments.

11. Autonomous Vehicles

Example: Self-driving cars using cameras and sensors to understand the road, detect
obstacles, and navigate safely.

Application: Improving road safety and enabling autonomous transportation.

12. Industrial Inspection

Example: Automated inspection of products on a production line to detect defects.

Application: Quality control in manufacturing, reducing human error and improving

efficiency.

11) What are the components of Conversational Al ? How it will work together to understand
and respond to human conversation?

Ans. Conversational AI systems consist of several key components that work together to
understand and respond to human conversations. These components integrate advanced AI
techniques, including natural language processing (NLP), machine learning, and sometimes
even computer vision. Here are the main components and how they interact:

1. Natural Language Understanding (NLU): Converts human language into a machine-

readable format. It involves tasks such as intent recognition, entity extraction, and context
understanding.

Example: If a user says, "Book a flight to Paris," the NLU identifies the intent ("book a flight")
and extracts relevant entities ("Paris").

2. Dialogue Management: Manages the flow of the conversation. Decides what the system
should do next based on the user’s input and the conversation context.

Example: Determines that after identifying the destination as Paris, the system should ask
for the date and time of the flight.

3. Natural Language Generation (NLG): Converts the system's responses into human-
readable language.

Example: Transforms the response data into a coherent and contextually appropriate
response, such as "When would you like to fly to Paris?"

4. Speech Recognition (if applicable): Converts spoken language into text. Often used in
voice-based conversational AI systems.

Example: Converts "Hey Siri, what's the weather today?" into text for further processing.

5. Text-to-Speech (TTS) (if applicable): Converts text responses back into spoken
language.

Example: Converts the textual response "The weather today is sunny with a high of 75
degrees" into audible speech.

6. Context Management: Keeps track of the conversation’s context and history to provide
coherent and contextually relevant responses.

Example: Remembers that the user has previously booked flights to Paris and might use this
information to offer personalized suggestions.

7. Knowledge Base/Database: Stores information and data that the conversational AI can
query to provide accurate responses.

Example: A database of flight schedules, prices, and user preferences.

8. Machine Learning Models: Enhance the system's ability to understand and generate
natural language. These models are often trained on large datasets to improve accuracy and
relevance.

Example: A machine learning model trained on millions of booking inquiries to accurately

predict user intents and preferences.

9. Integration Layer: Connects the conversational AI system with external systems and
APIs to retrieve or send data.

Example: Integrates with a flight booking system, a weather API, or customer relationship
management (CRM) software.
How These Components Work Together:

1. Input Processing: The user’s input, whether text or speech, is first processed by the
speech recognition component (if voice input) and then passed to the NLU
component.
2. Understanding the Input: The NLU component analyzes the input to determine the
intent and extract relevant entities.
3. Contextual Handling: The dialogue management component, with the help of
context management, decides the next action based on the current state of the
conversation.
4. Generating a Response: The NLG component creates a human-readable response.
5. Response Delivery: If the interaction is voice-based, the response is converted to
speech using TTS. If it is text-based, the response is sent as text.
6. Database Interaction: Throughout the process, the system may query the
knowledge base or databases to fetch relevant information or update records.
7. Learning and Improving: Machine learning models continuously learn from
interactions to improve understanding and response generation.

12) Explain the 1. Map Visualizations

2. Data Visualisation using charts and tables.

Ans. 1. Map Visualisation

Map Visualizations take on a new dimension. AI can leverage its capabilities to create more
advanced and informative map visualizations, transforming them from static displays to
dynamic tools for analysis and exploration.

AI algorithms can create customized maps based on user needs. Imagine a map that adjusts
traffic flow information based on real-time data or a map that dynamically highlights areas of
interest relevant to a specific search query.

AI can ingest data from various sources, including sensor networks, social media, and
satellite imagery, automatically populating maps with relevant information. This eliminates
manual data entry and ensures real-time updates.

2. Data Visualisation using charts and tables (in Power BI)

Data visualization is a crucial aspect of working with data in Power BI. It allows you to
transform raw data into easily understandable visuals that can reveal trends, patterns, and
relationships within your data set. Power BI offers a wide range of charts and tables to suit
your data analysis needs.

Charts:

 Bar Charts: Ideal for comparing categories. Each category is represented by a

rectangular bar whose length corresponds to the value of the measure being analyzed.
Great for visualizing things like sales figures by product category, customer count by
region, or website traffic by source.
 Column Charts: Similar to bar charts, but the bars are displayed horizontally. Useful
for situations where you have many categories or limited space on the report.
 Line Charts: Excellent for showing trends over time. The data points are connected
by lines, allowing you to see how a measure changes over a specific period. Useful
for visualizing sales trends over months, stock prices over time, or website traffic
patterns.
  Pie Charts: Best for representing proportional breakdowns of a whole. Each slice of
the pie represents a category, and the size of the slice corresponds to its percentage of
the total value. Effective for showing things like market share by competitor, budget
allocation by department, or customer demographics.
 Scatter Plots: Useful for exploring relationships between two numerical variables.
Each data point represents a single observation, plotted along horizontal and vertical
axes based on the two variables. Helpful for identifying correlations or clusters
between variables like customer age and purchase amount, product price and sales
volume, or employee experience and salary.
 Other Charts: Power BI offers additional charts like histograms, maps, and combo
charts (combining different chart types) to cater to specific data analysis requirements.

Tables:

 Simple Tables: The most basic visualization, displaying data in rows and columns.
Each cell contains a data point, and you can sort and filter the table to focus on
specific aspects of your data. Useful for presenting raw data, detailed breakdowns, or
comparisons across multiple categories.
 Matrix Tables: Offer a more compact way to display a large number of data points.
They allow you to group and summarize data by rows and columns, providing a
hierarchical view of your information. Helpful for analyzing complex data sets with
multiple dimensions, like sales figures by product category, year, and region.

Choosing the Right Visualization:

The best chart or table for your data depends on the type of information you want to convey.
Here are some general guidelines:

 For trends and comparisons over time: Line charts.

 For comparing categories: Bar charts, column charts, or pie charts (if you have a
limited number of categories).
 For relationships between two numerical variables: Scatter plots.
 For raw data exploration and detailed breakdowns: Tables.

13) Why should do data analysis with Power BI? Explain with Example.

Ans. Here's why you should consider data analysis with Power BI, illustrated with specific
advantages and examples:

Advantages of Power BI for Data Analysis:

1. Ease of Use: Power BI offers a user-friendly interface with drag-and-drop

functionality. This makes it accessible to a wider range of users, even those without
extensive data analysis experience.

Example: Imagine you're a marketing manager with limited technical expertise. You
can easily import customer survey data from a CRM system into Power BI and use
drag-and-drop features to create charts analyzing customer satisfaction by product
category. This empowers you to gain valuable insights without relying on IT
specialists.

2. Data Connectivity: Power BI connects to a vast array of data sources, including

Excel spreadsheets, cloud databases, and social media platforms. This allows you to
consolidate data from various sources into a single platform for comprehensive
analysis.
 Example: A sales manager can connect to data from the company's sales CRM,
warehouse inventory system, and marketing automation platform within Power BI.
This unified view enables them to analyze sales trends, identify correlations between
marketing campaigns and sales figures, and optimize inventory management based on
real-time data.

3. Visualizations Galore: Power BI boasts a rich library of charts and graphs, enabling
you to transform raw data into clear and informative visuals. This helps you identify
trends, patterns, and relationships within your data that might be missed in
spreadsheets.

Example: A financial analyst can create interactive bar charts in Power BI to

visualize budget allocations across different departments. By hovering over specific
bars, they can see detailed breakdowns of spending categories. This allows for a
clearer understanding of how the budget is being utilized and facilitates informed
decisions about resource allocation.

4. Empowering Self-Service BI: Power BI fosters a self-service BI (Business

Intelligence) environment. Business users can explore and analyze data
independently, without relying solely on IT departments. This speeds up the decision-
making process and fosters data-driven insights across the organization.

Example: A product manager can leverage Power BI to analyze customer reviews

and social media sentiment about a new product launch. They can identify areas for
improvement and make data-driven recommendations to the development team
without waiting for lengthy IT reports. This agility allows for quicker product
iterations and increased customer satisfaction.

5. Collaboration and Sharing: Power BI facilitates collaboration by allowing you to

share reports and dashboards with colleagues. This promotes teamwork and ensures
everyone is on the same page when it comes to data-driven decision making.

Example: A sales team can share a Power BI dashboard with regional managers,
highlighting sales performance metrics by territory. This collaborative approach
allows for better communication, enables managers to tailor sales strategies to specific
regions, and ultimately drive higher sales growth.

By incorporating these advantages and following the example scenarios, Power BI empowers
various users within an organization to leverage data for better decision making and
improved business outcomes.

14) Explain the Key Influences in Power BI.

Ans. The Key Influencers visual in Power BI can help identify important factors in your data
by analyzing the relationships between different variables and identifying which variables are
most strongly associated with a particular outcome. This visual provides a way to understand
the key drivers of a particular metric or outcome and can be used to support decision-making,
such as identifying which marketing campaigns are most effective in driving sales.

Once the Key Influencers visual is generated, it will display a list of the most important
variables that are influencing the target variable. The visual uses machine learning algorithms
to analyze the data and identify the relationships between variables. The Key Influencers
visually display the results as a list of cards, each representing an explanatory variable, with a
score indicating the strength of the relationship. The visual also displays a chart that shows
the relationship between the target variable and each explanatory variable.
By using the Key Influencers visual, you can identify which variables are most strongly
associated with a particular outcome and use this information to make data-driven decisions.

Example: Let’s imagine we’re working with a dataset of an online retail company that wants
to understand its customers better and identify the key factors influencing customer
satisfaction. We’ll use the Key Influencers visual to analyze the data and unveil those
factors.

Here’s a step-by-step guide on how to use the Key Influencers feature in Power BI:

Step 1: Create a new Power BI report

Open Power BI Desktop and create a new report. You can do this by clicking on ‘File’ >
‘New’ or simply starting Power BI Desktop for the first time.

Step 2: Load your data

To load your dataset into Power BI, click on ‘Home’ > ‘Get Data’ > ‘Excel/CSV/Text file’ or
select your preferred data source. Locate your dataset file and click ‘Open’. Once your data is
loaded, it will appear in the ‘Fields’ pane on the right side of the screen.

Step 3: Add the Key Influencer’s visual

In the ‘Visualizations’ pane, click on the Key Influencers icon (it looks like a small person
next to a bar chart) to add a new Key Influencers visual to your report canvas. You can also
use the search bar to find the visual by typing ‘Key Influencers.’

Step 4: Configure the visual

Now, let’s configure the Key Influencers visual to analyze customer satisfaction. First, drag
the ‘Customer Satisfaction’ field from your dataset into the ‘Analyze’ section. Let’s say we
have a column in our dataset called ‘Customer Satisfaction’ with values ranging from 1 (low
satisfaction) to 5 (high satisfaction).

Next, drag the fields you want to explore as potential influencers into the ‘Explain by’
section. For example, we might want to see how factors like ‘Product Category,’ ‘Delivery
Time,’ ‘Discount,’ and ‘Customer Age’ influence satisfaction.

Step 5: Interpret the results

The Key Influencers visual will display the top factors influencing customer satisfaction in
your data, along with a ranked list of associated values or categories.

For instance, it might show that customer satisfaction is highly driven by ‘Delivery Time’
and ‘Product Category.’ To dive deeper, it might reveal that faster delivery times (e.g., 1-2
days) lead to higher satisfaction scores.

Further down the visual, you’ll also see the ‘Top Segments’ section, which can help you
better understand the characteristics of the dataset. For example, it might show that a
significant portion of your customers are aged 25-35 and prefer ‘Electronics’ products.

Step 6: Customize and explore

You can customize the Key Influencers visual by clicking on the paintbrush icon and
adjusting styles, colors and other formatting options. Feel free to continue exploring the data
by adding or removing fields in the ‘Explain by’ section or using filters to refine your
analysis.