Machine

Learning
Table of contents

01 What is Machine Learning?

02 Types of Machine Learning

03 Applications in Various Fields

04 Limitations & Future Directions

01
What is Machine
Learning?
Machine Learning
Machine learning is a branch of artificial intelligence
that focuses on developing algorithms and techniques
that enable computers to learn from data and
improve their performance over time without being
explicitly programmed. It involves training models on
labeled or unlabeled data, extracting patterns, and
making predictions or taking actions based on those
patterns.
Artificial
intelligence Ability of Machine to
imitate Human Intelligence

Machine Algorithms to incorporate

learning intelligence into machine
by automatically learning
from data

Deep Algorithms that mimics

learning human brain to incorporate
intelligence into machine
 Difference between traditional programming and
machine learning
Traditional Programming Machine Learning

Ø Instructions and rules provided Ø Data-driven approach

explicitly
Ø Relies on predefined rules and Ø Relies on data
logic
Ø Less adaptable to new scenarios Ø Can adapt to new situations by
adjusting internal parameters
Ø Performs within defined Ø Can generalize patterns for new,
boundaries unseen data
History of Machine Learning
Machine learning evolved within artificial intelligence. Early work focused on symbolic approaches and
expert systems. Decision trees and neural networks emerged as powerful tools. The availability of large
datasets and increased computational power led to significant advancements. Deep learning, with deep
neural networks, revolutionized tasks like image recognition. Reinforcement learning made strides in game
playing and control systems. Ongoing research explores explain ability, fairness, and ethical considerations.
Machine learning integrates with other fields for diverse applications.
 Types of Data
Structured Data Unstructured Data Semi-Structured Data

Organized and formatted Lacks a predefined Has some level of

in a consistent tabular structure and is in a organization but does not

structure, with rows and natural, human-readable adhere to a strict tabular

columns. Examples format. Examples include structure. Contains partial

include data in databases text documents, images, organization or metadata.

and spreadsheets. audio files, and social

media posts.
02
Types of Machine
Learning
 Machine Learning Types

Supervised Learning 1

Unsupervised Learning 2

Reinforcement Learning 3

Deep Learning 4
Supervised Learning
Supervised learning is a type of machine learning in which the algorithm learns from
labeled examples to make predictions or classify new, unseen data. In supervised
learning, the training data consists of input features (also known as independent
variables) and corresponding target variables (also known as dependent variables or
labels). The goal is to learn a mapping between the input features and the target
variables, allowing the algorithm to generalize and make predictions on new, unseen
data.
Types of supervised learning
Classification

Classification is defined as the process

of recognition, understanding, and
grouping of objects and ideas into
preset categories a.k.a “sub-
populations.” With the help of these
pre-categorized training datasets,
classification in machine learning
programs leverage a wide range of
algorithms to classify future datasets
into respective and relevant
categories.
 Types of supervised learning
Regression

It is a method to determine the statistical

relationship between a dependent variable
and one or more independent variables.
The change in the independent variable is
linked to altering the dependent variables.
 Linear Regression
AI Linear Regression is a supervised machine
learning algorithm with a continuous and
constant slope expected performance.
Rather than attempting to classify values
into groups (such as cats and dogs), it is used
to estimate values within a continuous range
(such as price and revenue).
 Support Vector Machine

The goal of the SVM algorithm is to create

the best line or decision boundary that can
segregate n-dimensional space into classes
so that we can easily put the new data point
in the correct category in the future. This
best decision boundary is called a
hyperplane.
 Bear

Cat

Cat
Unsupervised
Learning
In Unsupervised Learning, the machine uses
unlabeled data and learns on itself without any
supervision. The machine tries to find a pattern in the
unlabeled data and gives a response. Let's take a
similar example is before, but this time we do not tell
theCREDITS:
machine This presentation
whether it's atemplate
spoon orwas
a created by Slidesgo
knife. The
and includes icons by Flaticon, infographics & images by
Freepik identifies
and contentpatterns
by Eliana from
Delacour
machine the given set and
groups them based on their patterns, similarities, etc.
Types of unsupervised learning

Clustering Association
 Market Basket Customer

Analysis: Segmentation:
By analyzing transaction data,
unsupervised learning Unsupervised learning can

algorithms can identify be used to cluster customers

frequently co-occurring items based on their purchasing

and uncover associations or behavior, demographics, or

patterns. This is useful for other relevant data. This

cross-selling or helps businesses understand

recommendation systems.
Reinforcement Learning
Reinforcement learning is a powerful type of machine
learning where an agent learns to interact with an
environment and improve its performance through trial
and error. It is inspired by how humans and animals learn
from their experiences. In this process, the agent
receives feedback from the environment in the form of
rewards or penalties, guiding its learning journey.
Ø Four points
§ Agent
§ Environment
§ Action
§ Reward

Ø Applications of Reinforcement

§ Playing
§ Robotics
§ Autonomous Systems

Ø Summary
 Deep Learning
Deep learning is a subset of machine learning that focuses
on training artificial neural networks with multiple layers,
Neural network
also known as deep neural networks. It has revolutionized
many fields, including computer vision, natural language
processing, and speech recognition, by enabling the
development of highly accurate and complex models.

Ø Inspired by Human brain

Ø Architecture of Neural Network
Ø Layers
§ Input Layer
§ Hidden Layers
§ Output Layers
Ø Architectures

§ Recurrent Neural Networks (RNN)

§ Convolutional Neural Networks (CNN)
§ Generative Adversarial Networks (GAN)

ØSummary
Training & Evaluation
Ø Three Aspects
§ Splitting Data
§ The Training Process
§ Model evaluation metrics

Ø Split into Three Sets

§ The Training Set
§ Validation Set
§ Test

Ø Summary
 Model Deployment
Deployment is the method by which you integrate
a machine learning model into an existing production
environment to make practical business decisions
based on data. It is one of the last stages in
the machine learning life cycle and can be one of the
most cumbersome. Often, an organization’s IT
systems are incompatible with traditional model-
building languages, forcing data scientists and
programmers to spend valuable time and
brainpower rewriting them.
 Considerations and challenges

Deployment
Scalability 01 02 environments

03 Versioning and
Performance 04
model governance

Security and 06
05 User acceptance
privacy
Ethical Considerations

Bias and Fairness

Transparency and
Explainability

Responsible AI
Practices

Data Privacy and

Security
03
Real Life Applications
 Healthcare
Machine learning has been
instrumental in healthcare applications,
such as disease diagnosis and drug
discovery. ML algorithms can analyse
patient data, medical imaging, and
genetic information to assist in
accurate diagnosis and develop
targeted treatment plans. Additionally,
machine learning models can aid in
drug discovery by analysing vast
amounts of biological data to identify
potential candidates for new
medications.
 Finance
Machine learning plays a significant role in the
finance industry, particularly in fraud detection and
algorithmic trading. ML algorithms can analyse large
volumes of financial transactions in real-time to
identify fraudulent patterns and prevent
unauthorised activities. In algorithmic trading,
machine learning models can analyse market data,
predict trends, and make informed trading decisions
to optimise investment strategies.
 Marketing
Machine learning techniques enable
marketers to make data-driven decisions
and improve customer engagement. ML
models can analyse customer behaviour,
preferences, and demographics to
segment customers effectively. This
segmentation helps in delivering
personalised recommendations, targeted
advertisements, and optimising marketing
campaigns for better customer satisfaction
and higher conversion rates.
04
Limitations & Future
Directions
Limitations of Machine learning

Data Limited
01 04 Contextual
Dependency
Understanding

Overfitting and Adversarial

02 05
Underfitting Attacks

Lack of Common
Lack of
03 06 Sense
Explainability Reasoning
 Federated Learning

Federated learning is a distributed

approach to machine learning
where the training process is
performed on decentralised edge
devices or servers, instead of a
centralised data repository.
 Edge Computing

Edge computing is a computing

paradigm that brings computation and
data storage closer to the edge of the
network, near the data source or the
end user, rather than relying solely on a
centralised cloud infrastructure.
 01 02
Federated Learning
ML in Healthcare for Privacy

03 04

Edge ML Emerging Trends Ethical ML

05 06
ML in Autonomous
ML in Natural
Systems
Language Processing
 Importance of Machine Learning

Machine learning is of paramount importance as it empowers organisations to unlock valuable

insights, make data-driven decisions, and automate complex tasks. By leveraging algorithms
and statistical models, machine learning enables the extraction of patterns, predictions, and
correlations from large datasets, leading to enhanced efficiency, accuracy, and innovation. Its
applications span across various domains, including healthcare, finance, marketing, and more,
revolutionising industries and driving advancements. In an era where data is abundant, machine
learning serves as a catalyst for progress, enabling organisations to harness the power of
artificial intelligence and unleash its transformative potential.
Conclusion
It's important to note that
ongoing research and
advancements in the field
of machine learning are
continuously addressing
these limitations to improve
the effectiveness,
interpretability, and
fairness of models.
 Any
Questions?
Thanks!