Mujahid Sultan
AIP Tech Trends
Machine Learning
The most amazing invention
of the 20th Century was ?
The most amazing invention
of the 20th Century was ?
COMPUTER
In 21st Century
we will reach
new heights in
technology
which may endanger our
very existence ?
Machine Learning is the basis of Artificial intelligence
How can we make machines learn.
(with a simple example)
Learning Objectives are:
Machine Learning
What is
machine ?
Some Basics
Vector (x, y)
Vector 1 (x1, y1)
Vector 2 (x2, y2)
Vector 1 (x1, y1)
Vector 2 (x2, y2)
Can a machine calculate distance between these?
Vector 1 (x1, y1)
Vector 2 (x2, y2)
Can a machine calculate distance between these?
(0,0)
Vector 1 (x1, y1)
Vector 2 (x2, y2)
(x2-x1)
(y2-y1)
(0,0)
Can a machine calculate distance between these?
Vector 1 (x1, y1)
Vector 2 (x2, y2)
(x2-x1)
(y2-y1)
d2 = (x2-x1)2 + (y2-y1)2)
(0,0)
Distance between vectors
Vector 1 (a1, b1, ………,n1 )
Vector 2 (a2, b2, ……….., n2)
d2 = (a2-a1)2 + (b2-b1)2 + ……+ + (n2-n1)2)
In higher dimensions:
Vector 1 (a1, b1, ………,n1 )
Vector 2 (a2, b2, ……….., n2)
Euclidian Distance
In higher dimensions:
Vector 1 (a1, b1, ………,n1 )
Vector 2 (a2, b2, ……….., n2)
Euclidian Distance
In higher dimensions:
Any other distance measures?
Vector 1 (a1, b1, ………,n1 )
Vector 2 (a2, b2, ……….., n2)
Correlation
In higher dimensions:
Vector 1 (a1, b1, ………,n1 )
Vector 2 (a2, b2, ……….., n2)
Manhattan Distance
In higher dimensions:
Vector 1 (Normal)
Vector 2 (Cancerous)
Profile of human genome, high dimensional vector.
Vector 1 (Normal)
Vector 2 (Cancerous)
Vector 1 (Normal)
Vector 2 (Cancerous)
Vector 1 (Normal)
Vector 2 (Cancerous)
(Normal)
(Cancerous)
Data Set
(Normal)
(Cancerous)
Randomly select two points
in the (domain) space
1
2
(Normal)
(Cancerous)
2
1
Calculate distance from 1St point
to each data point
(Normal)
(Cancerous)
12
1
Calculate distance from 2nd Point
to each data point
(Normal)
(Cancerous)
Move these two points to the mean
of data points closer to each
1
2
(Normal)
(Cancerous)
Move these two points to the mean
of data points closer to each
2
1
(Normal)
(Cancerous)
Calculate distance from each point and
and move it to the mean of group
2
1
(Normal)
(Cancerous)
Calculate distance from each point and
and move it to the mean of group
2
1
(Normal)
(Cancerous)
Calculate distance from each point and
and move it to the mean of group
2
1
(Normal)
(Cancerous)
Calculate distance from each point and
and move it to the mean of group
Stop when the means are not changing
significantly
2
1
(Normal)
(Cancerous)
2
1
Calculate distance from each point and
and move it to the mean of group
Stop when the means are not changing
significantly
(Normal)
(Cancerous)
2
1
Calculate distance from each point and
and move it to the mean of group
Stop when the means are not changing
significantly
(Normal)
(Cancerous)
2
1
Calculate distance from each point and
and move it to the mean of group
Stop when the means are not changing
significantly
(Normal)
(Cancerous)
2
1
Calculate distance from each point and
and move it to the mean of group
Stop when the means are not changing
significantly
(Normal)
(Cancerous)
For any new patient,
machine can classify
2
1
Today we learned a simple example
how we can make machines learn,
called
Machine Learning which is the
basis of Artificial intelligence

Introduction to Machine Learning