Introduction to Machine Learning
Ketan Kotecha, PhD(IIT Bombay)
https://www.linkedin.com/in/ketankotecha/
What is Artificial intelligence?
Ability to learn, understand and think” -Oxford
Dictionary
Intelligence is also the ability to learn from the
environment and change our behavior based on
AI inputs we get. The study of how to make
computers perform functions which at present
humans are good at.
Examples: Voice and Speech Recognition, Face
recognition and face identification, Object
detection, Intuition, Inferencing, Learning new
skills, Decision making,
THE JOURNEY OF LEARNING SO FAR….
1950-1980 1980-2010 2010 onwards...
Artificial Intelligence
mimicking human
intelligence, using
logic, if else rules or
machine learning
Machine Learning
Improving at task
from Experience
Deep Learning
Learning from vast
amount of data
ML Vs DL
Types of Machine Learning
2
1
3
UNSUPERVISED
SUPERVISED Data Driven
REINFORCEMENT
Task Driven
Finds Clusters Learn from
Predicts Next Value
(Outcome is Labelled) mistakes/Achievments
(Outcome is Labelled)
Eg: Regression, Eg: Association,
Classification Clustering
Supervised Learning
Unsupervised Learning
Reinforcement Learning
AI has reached everywhere…
Supervised Learning
Types of Supervised
Learning
Classification 1 Supervised 2 Regression
Learning
In supervised learning, algorithm is selected based on target variable.
Pooja Kamat
Types of Supervised Learning (Contd.)
If target variable is categorical (classes), then use classification algorithm.
In other words, classification is applied when the output has finite and
motion discreet values.
Example: Predict the class of car given its features like horsepower,
Classification mileage, weight, colour, etc.
The classifier will build its attributes based on these features.
Analysis has three potential outcomes - Sedan, SUV, or Hatchback
Pooja Kamat
Types of Supervised Learning (Contd.)
If target variable is a continuous numeric variable (100–2000), then use a regression algorithm.
Example: Predict the price of a house given its sq. area, location,
no of bedrooms, etc.
A simple regression algorithm is given below
Regression
y=w*x+b
This shows relationship between price (y) and sq.
area (x)
where price is a number from a defined range.
Pooja Kamat
Types of Supervised Learning (Contd.)
Classification Regression
What Class How much ?
?
Pooja Kamat
Supervised Learning Flow
𝑓𝑥
Training and 70%
Testing
Training Machine Statistical
Dataset Learning Model
Historical Random
Data Sampling
Prediction 30%
Test Prediction and
Dataset Testing
Model Validation Outcome
Pooja Kamat Original Source: Simplilearn.com
Breast Cancer (Malignant , Benign)
Machine Learning Specialization by Andrew Ng
Regression
Housing Price Prediction
Understanding the Algorithm
Input Output (Labels)
Train the algorithm
x1 y1
(reverse-engineer the
relationship between x
x2 y2 and y)
x3 y3 Learned
algorithm
x4 y4
y = f(x)
Training x5 y5 (y is some function of x)
Data
x6 y6
x7 y7
Predict y for the new
x8 y8 instance x new using y = f(x)
x9 y9
x10 y10 Original Source: Simplilearn.com
Pooja Kamat
Supervised Learning
Classification Vs Regression
Unsupervised Learning
Supervised Vs Unsupervised
Machine Learning Specialization by Andrew Ng
Thank you!
Questions
Ketan Kotecha, PhD(IIT Bombay)
Symbiosis Centre for Applied Artificial Intelligence (SCAAI)
Symbiosis International (Deemed University)
