This lecture introduces supervised learning, a machine learning paradigm that uses labeled datasets to train models for predicting outputs on new, unseen data. It covers the fundamental concepts, components, a key type (classification), and the typical workflow involved in supervised learning problems.
Mind Map
Click to expand
Click to explore the full interactive mind map • Zoom, pan, and navigate
Dear students, today our lecture is
about supervised learning which is part
First we will look at the overview of
the lecture.
In the start we will recap the our last
lecture which is types of machine
learning. Then we will look at the
objectives of today's lectures. What is
supervised learning today's main topic?
What are the different components of
supervised learning? Then we will look
at a type of supervised learning which
is classification.
And then we will see what are the
different steps involved in a supervised
learning problem.
And then we will see what we'll uh study
in the next lecture
and at the end we will look at the
lecture sum summary and then you will
have a chance to ask questions.
So what are the learning objectives of
the today's class?
Explain the concept of supervised
learning and its core components.
Understand the concept of classification
which is a type of uh supervised learning.
learning.
Identify common real life applications
of supervised learning and describe the
basic workflow of supervised learning
model. [clears throat]
So in the last lecture we studied about
what are the different types of machine
learning which is supervised learning
unsupervised learning and re
reinforcement learning and to today we
will specifically focus on supervised learning.
So supervised learning is a type of
machine learning where we need to have
the label data sets. If we don't have
label data sets then we cannot use
supervised learning. In that case we
have a different type of machine
learning which is called unsupervised
learning. We can use that.
So what is the objective of supervised
learning? Predict outputs labels from
new unseen data. So the model is trained
on uh on data which has which already
has labels. These labels are provided by
some experts in their domain depending
upon what type of data we have. And then
we provide new data which has previously
not seen by the model and it tries to predict
predict
the class or specific number based on
what type of super basically supervised
machine learning has different types
that we will study later on. So there
are some examples here real life
examples where supervised learning can
be used is for example when we get
emails they can be spam or normal emails
and you normally use uh for example any
email for example Gmail you know that
some of your emails goes into spam
folder and the others that are normal
comes to your email right so what what
is happening
behind that. So they are actually using
a machine learning algorithm which has
learned based on data that what kind of
data is what kind of emails are spam and
what are not spam right so this is
basically classification and then we
have another type of machine uh
supervised machine learning uh where we
try to find a continuous value. So if
you if you see this example we have
different uh type of houses they have
different size of uh they have different
number of rooms and this is the price
how much. So if for example based on
data if you want to predict another
house uh for example which has four
rooms and we want to predict the price
so at at what price we should sell our
house. So we can use regression. In
regression we find a single continuous
value. Okay. So now that you understand
uh uh what we are doing here in
supervised learning can can you give me
another example
of supervised machine learning where
where we can use supervised machine
learning for for example classification
or regression.
Nice sir you are right. So we can use
them in those problems as you mentioned.
Let's move on.
So what are the key components of
uh supervised machine learning when you
have data right? So these are parameters
for for example these are uh independent
variables and then we want to find the
price of the house. So that value is
dependent on these values. So we say it
is a dependent where y is a dependent
variable on x. We have x1, x2, x3. We
have different properties of the of the
house, different fields and based on
that we have a price. So these prices we
call it label or we call it dependent
variable. Right? So this is also know
known as a feature vector right and if
we don't have these labels we won't be
able to use supervised learning because
we have to train our model that if we
have these parameters then this is the
output based on those values. So if we
don't have
these values we can cannot use
supervised learning. So that is the main
point here right which is also called
label right.
Okay. Now we look at a specific type of
supervised learning which is called classification.
classification.
So what is basically what happens here
that we we have to uh divide our data or
categorize our data into groups. For
that purpose we use classification. Our
model has to learn a slope here to
classify. In this case, we are
separating the circles and squares. This
is called binary classification because
we are dividing it into two classes. And
here it's a multiclass problem. We are
dividing it into
into three groups. So it's multiclass classification.
So what is the general general workflow
of uh um supervised machine learning
problem? When you have data consider
that our data is clean and it's ready.
So what we will do we divide the data
into training set and test set. Remember
in training set we also have labels but
in test test set we don't have labels.
Right? Here we have labels but here we
don't have the labels. If we have them
we have to remove them. Right? Then we
train our model on the training data.
Once our model is trained we test our
model based on the test data which
doesn't have the labels. So this is
trained because we have labels here. We
don't have labels and it has to tell us
the labels. Right? labor is like the
answer which class it belong and then at
the end we have the result. So the train
model make predictions on the test data.
So this is basic workflow of supervised learning.
So in today's lecture
we we studied what is supervised
learning what are the different
components we studied a type of
supervised learning which is classification
classification
and then we looked at the steps involved
in supervised learning workflow.
In the next lecture we will study uh
another type of
um supervised learning which is
regression where we find a value.
At this point, I give chance to my
students to ask questions. So, here they
can ask questions. Uh, and my philosophy
is that we should interact with students
continuously, ask questions from them
and also allow them to ask questions.
uh and it needs to be a comfortable
environment in the classroom where they
can involve
uh in the discussion so that they
understand it better. And my another uh
philosophy is that we need to tell the practical
practical
applications of all the theoretical
Click on any text or timestamp to jump to that moment in the video
Share:
Most transcripts ready in under 5 seconds
One-Click Copy125+ LanguagesSearch ContentJump to Timestamps
Paste YouTube URL
Enter any YouTube video link to get the full transcript
Transcript Extraction Form
Most transcripts ready in under 5 seconds
Get Our Chrome Extension
Get transcripts instantly without leaving YouTube. Install our Chrome extension for one-click access to any video's transcript directly on the watch page.
