PART-B: SUBJECT SPECIFIC SKILLS

UNIT-1: AI Reflection, Project Cycle and Ethics

Artificial Intelligence (AI)

Definition:
Artificial intelligence is a technology that enables computers and machines to
simulate human intelligence and problem-solving capabilities.
Components of AI:
Data and Algorithm makes a machine Artificially intelligent.
Features of AI:
 Mimics human intelligence
 Solves real-world problems
 Improves on its own from past experiences
 Predicts and make decisions on its own
Applications of AI:
 Face Lock and Fingerprint Lock in Smartphones protects user data.
 Smart assistants like Apple’s Siri and Amazon’s Alexa recognize patterns
in speech, then infer meaning and provide a useful response.
 Fraud and Risk Detection by Banks and Finance companies through
customer profiling, past expenditures, and other essential variables to analyse
the probabilities of risk and default.
 Medical Imaging helps to understand patient’s health condition and diseases.
Machine Learning (ML):
Machine learning (ML) is a branch of AI that focuses on the using data and
algorithms to enable AI to imitate the human skills of learning and gradually
improving its accuracy. Machine learning may dependent on human
intervention to learn.
Deep Learning (DL):
Deep learning is a subset of ML that uses multi-layered neural networks to
simulate the complex decision-making power of the human brain.
Deep Learning does not need human intervention.

KVS ZIET MYSORE 66

 Domains of AI
There are 3 major domains in AI:

Computer Natural Language

Data Science
Vision Processing

1. Data Sciences (DS)

Data sciences is a domain of AI related to data systems and processes where the
system collects numerous data, maintains data sets and derives meaning or sense
out of them. The information extracted through data science can be used to make
a decision about it.

KVS ZIET MYSORE 67

 Application of Data Science:
Price Comparison Websites like PriceGrabber, PriceRunner, Junglee, Shopzilla,
DealTime etc.

2. Computer Vision (CV)

Computer Vision is a domain of AI that enables a machine to get and analyse visual
information and predicts or decides about it.
The entire process involves image acquiring, screening, analysing, identifying and
extracting information from photographs and videos captured through thermal or
infrared sensors, indicators or other sources.
Application of Computer Vision:
Self-Driving cars/ Automatic Cars, Face Lock in Smartphones, etc.

3. Natural Language Processing (NLP)

Natural Language Processing is a branch of AI that deals with the interaction
between computers and humans using the natural language. NLP attempts toextract
information from the spoken and written words using algorithms.
The objective of NLP is to read, decipher, understand, and make sense of the human
languages.

Application of Natural Language Processing:

Email filters / Spam filters, Smart assistants like Alexa, Siri etc.
AI Ethics
Difference between Morals and Ethics:

KVS ZIET MYSORE 68

 AI Ethics:
Ethical principles need to be applied by the organisations developing and
implementing AI Systems regarding the volume and types of data need to be
collected from the users in order to provide optimal customised services. The
privacy and secrecy of the users needs to be maintained.
Principles in AI Ethics:

Human
Rights

Inclusion Ethical Bias

Principles

Privacy

 Human Rights: The users of AI Systems should not loose personal and financial
independence. The must not face any discrimination.

 Bias: Bias is said to be the partiality or preference for one over others. AI bias
may often come from the collected data which may affect the results. For
example, AI should represent all sections of the society and must not exclude
anyone.
 Privacy: AI Systems should not disclose individual and private data. It should
not breach one’s safety and security.
 Inclusion: AI must discriminate against a particular group of population, causing
them any kind of disadvantage.

Ethical scenarios faced while building AI solutions

Scenario 1: If a driverless car finds someone crossing the road, takes a turn to
avoid hitting the person and instead smashed the car in a wall / tree nearby

KVS ZIET MYSORE 69

 causing serious injury to the passengers, is it a fault in the development on moral
ground.
Scenario 2: If any Automated or Driverless Car hits someone standing in the
middle of a road, who needs to be held responsible; whom should be penalised.

AI bias
The biased mentality of the developers may result in the presence of bias in the
AI System. Bias might not be negative always but it may reflect the social norms
and common facts.
Examples of AI Bias:
 All the virtual assistants have a female voice.
 If searching is done for Salons on Google, the resultant list mostly contains
female salons.
 Searching for images of Nurse gives the pictures of female nurses mostly.
AI Access
Since AI Systems are emerging technology and expensive, all cannot get access
to AI and get benefitted. AI may create unemployment if not implementedcarefully
and with compassion.
Access of AI to kids must be limited for their effective cognitive development.
Smart Assistants can be used for fun, but problem-solving software like ChatGPT
should not be used for solving math and science related problems.

AI Project Cycle
Project Cycle is a step-by-step process to solve problems using AI and drawing
inferences about them.
Importance of the AI project cycle:
Understandability: Understand the process and its each step better
Modularity: Break down process into small parts for ease of development.
Efficiency: Make better, faster and more effective AI solution in less effort

KVS ZIET MYSORE 70

 Fig: AI Project Cycle

Stage 1: Problem Scoping

Selection of a problem to solve with the help of AI is said to be Problem Scoping.
Step 1: Selection of theme.
Step 2: Find root cause of the problem.
Step 3: Select GOAL of the AI project.
Step 4: Identify 4 critical parameters using 4Ws Problem Canvas to solve.
Step 5: Problem Statement template can be used to frame the 4Ws to define the
problem and summarize the key elements.

Who – Stakeholders of the Problem – Person or people facing the problem.

What –
I. Identify the nature of the problem
II. Assess whether it is actually a problem to the stakeholders
III. Evidence in support of the problem

KVS ZIET MYSORE 71

 Where – Finding the following details about the problem:
Location, Context, Situation, Frequency
Why – Finding root cause of the problem.

Stage 2: Data Acquisition

Data are the piece of facts and information collected for reference or analysis.
Data must be accurate, reliable and traceable.
Data acquisition refers to collection of correct data from reliable source.
AI system must be trained with Training Data Set. Based on Historical / Training
data, the AI system can predict next level of data set which is called Testing Data.
Data can be acquired in the various formats – Coordinates, texts, photos, videos
etc. Data feature refers to the type of data i.e., attributes / fields.
Data acquisition methods:
Surveys Web Scraping Sensors Cameras Observations APIs

Only public data should be acquired that is available in open-source websites

or government portals. General Data Protection Right (GDPR) must not be
violated as it’s punishable offense. Example of such website: data.gov.in,
india.gov.in, etc.
Stage 3: Data Exploration
Visual representation of data helps in analyzing data in the following ways:
 Quickly comprehend the trends, relationships, and patterns within data
 Define strategy and choose model
 Communicate information to others

Different types of graphs can be used: Line graph, Bar Graph, Pie chart etc.

Stage 4: Data Modelling

For developing project in AI the model or algorithm can be chosen to predict

output based on the set of inputs in any one of the following manners:
KVS ZIET MYSORE 72
  Designing own model
 Choose a pre-existing model

AI Models

Rule-based Learning-based
Approach Approach

Machine Learning Deep Learning

(ML) (DL)

A. Rule-based AI Approach:
Relationships or patterns in data are predefined and algorithm follows the rule
or instructions mentioned and perform the tasks accordingly. The data along with
rules are fed to the machine during training and then the machine become able to
predict answers for the same.
Example:
Step 1: Train system with training data fed into the system.
[Dataset containing 1000 images of onions and carrots with labels]
Step 2: Feed a testing data [Say one image of onion]
Step 3: Compare training data with testing data as per rules
[Compare image of onion with all others]
Step 4: Identify the correct output [Determine its onion]
Advantage:
The algorithms are simple and easy to implement. Number of data required is
limited. Hence training machine is easy.
Limitation:
This learning is static. The machine once trained, does not take into consideration
any changes made in the original training dataset. That is, if you try testing the
machine on a dataset which is different from the rules and data you fed it at the
training stage, the machine will fail and will not learn from its mistake. Once
trained, the model cannot improvise itself based on feedbacks.

KVS ZIET MYSORE 73

 B. Learning-based AI Approach:

In Learning based AI model, the machine gets trained on the data fed to it and
then can design a model, adaptive to the change in data. Implementation through:
Classification of images, used in Computer Vision.
Advantage:
It is a dynamic model. If the model is trained with a type of data and the machine
designs the algorithm around it, the model would adjust itself according to the
changes in the data to handle the exceptions.

Disadvantage:
Huge amount quality of data is required for training the machine. Large storage
and efficient algorithm is required. It is expensive and time taking to implement.

Example:
Step 1: Random data is fed into the system. [10,000 images of people in a city]
Step 2: Machine analyses data. [to identify sick and healthy people]
Step 3: System tries to extract similar features. Algorithm needs to derive
relationship in data. [Identify facial expressions and emotions]
Step 4: Cluster same data together. [Form group with identical facial
expressions]
Step 5: Output is the broad trends observed in the data set. [Identify whether a
given picture belong to a sick or healthy person]

KVS ZIET MYSORE 74

 Supervised Learning
In a supervised learning model, the dataset which is fed to the machine is labelled.
A label is some information which can be used as a tag for data.
i. Regression:
Algorithm generates a mapping function from the given data, represented by a line.
It helps to predict or forecast future data. Regression works with continuous data.
e.g. – Prediction of marks in the next exam based on historical data.
ii. Classification:
Algorithm classifies the data according to the labels and sorted as per labelling is
done. It works on discrete data sets.
e.g. – Classify image of men and women where numerous images of men and
women in different structures and formats are fed as training data.

Unsupervised Learning
An unsupervised learning model works on unlabelled dataset. Data fed to the
machine is random. The unsupervised learning models are used to identify
relationships, patterns, and trends out of the training data. It helps the user in
understanding what the data is about and what are the major features identified by
the machine in it.
Example
A random data of 1000 dog images are fed into the system and some pattern can be
found out of it, like colour, size of dogs etc.
i. Clustering:
This unsupervised learning algorithm can cluster the unknown data according to the
patterns or trends identified out of it. Cluster works in random, unlabelled, and

KVS ZIET MYSORE 75

 discrete data sets. The patterns observed might be the ones which are known to the
developer or it might even come up with some unique patterns out of it.

ii. Dimensionality Reduction:

Human beings are able to visualise upto 3-Dimensions only. But according to lot
of theories and algorithms, there are various entities which exist beyond 3-
Dimensions. Dimensionality reduction algorithm is used to reduce dimensions and
still make sense out of data.
The information gets distorted with reducing dimensions. At least 50% of the
information is lost after reducing one dimension.

Reinforcement Learning:
Reinforcement learning is a machine learning technique that trains software to make
decisions to achieve the most optimal results. It mimics the trial-and-error learning
process that humans use to achieve their goals. This uses a reward-and- punishment
paradigm as they process data.

KVS ZIET MYSORE 76

 Stage 5: Evaluation
After completing all the steps of AI Project Cycle, the model needs to be evaluated to ensure
correctness of results generated from new data.
Once a model has been made and trained, it needs to go through proper testing so that one
can calculate the efficiency and performance of the model. Hence, the model is tested with
the help of Testing Data (acquired during Data Acquisition stage) and the efficiency of the
model is calculated based on the parametersmentioned below:

If there is any variance between Training data set and Test data set, the model needs to be
improved by means of iteration.
Receiver Operator Characteristics (ROC) is a metric used to find out the accuracy of a
model.

KVS ZIET MYSORE