Skip to player
Skip to main content
Skip to footer
Search
Connect
Watch fullscreen
Like
Comments
Bookmark
Share
Add to Playlist
Report
Lecture 07: A Feedback Loop of an AI Project Framework
Nafees AI Lab
Follow
5/10/2025
In this lecture, we will learn how can we find a problem. What is the significance of problem definition. A feedback loop will generate using an example of AI Project Framework.
Category
π
Learning
Transcript
Display full video transcript
00:00
Welcome back!
00:02
This lecture we will talk about machine learning framework.
00:06
Last lecture we have seen
00:08
a holistic view of machine learning data science
00:10
and then we have seen
00:12
supervised, unsupervised
00:14
all the categories we have seen. However,
00:16
if you remember,
00:18
I remember that data science
00:20
basically
00:22
3 components
00:24
1. Pre-processing
00:26
2.
00:28
We have a basic action
00:30
or processing
00:32
then
00:34
we have a post processing
00:36
3
00:38
things
00:40
I think
00:42
I will shift the names
00:44
so that we will
00:46
get into the context
00:48
Pre-processing
00:50
basically
00:52
data collection
00:54
processing
00:56
which is
00:58
modeling
01:00
basically
01:02
machine learning model
01:04
apply
01:06
and apply
01:08
different steps
01:10
and then
01:12
post processing
01:14
then
01:16
you can deploy
01:18
to
01:20
data science
01:22
basically
01:24
machine learning
01:26
project
01:28
three steps
01:30
okay
01:32
now
01:34
we have to
01:36
explore this
01:38
modeling
01:40
modeling
01:42
modeling
01:44
modeling
01:46
modeling
01:48
to
01:50
set it
01:52
to
01:54
set it
01:56
to
01:58
set it
02:00
to
02:02
set it
02:04
to
02:06
set it
02:08
to
02:10
resemble
02:12
now
02:14
we have to
02:16
see it
02:18
as
02:20
as
02:22
as
02:24
as
02:26
as
02:28
as
02:30
as
02:32
as
02:34
problem
02:36
definition
02:38
definition
02:40
you can see
02:42
that
02:44
problem
02:46
you can see
02:48
that
02:50
problem
02:52
you can see
02:54
solve
02:56
this
02:58
problem
03:00
you can see
03:02
you can see
03:04
basically
03:06
information
03:08
that
03:09
you can see
03:10
lack of information
03:11
or problem
03:12
or glitch
03:14
so
03:15
lack of information
03:16
what was
03:17
data
03:18
basically
03:19
you can see
03:20
that
03:21
the data
03:22
that
03:23
you can see
03:24
that
03:25
you can see
03:26
that
03:27
you can see
03:28
that
03:29
there is
03:30
a problem
03:31
there is
03:32
a problem
03:33
there is
03:34
a problem
03:35
there is
03:36
a problem
03:37
there is
03:38
a problem
03:39
you need to see
03:40
that
03:41
data
03:42
which data
03:43
is
03:44
which data
03:45
is
03:46
different types
03:47
of data
03:48
structured
03:49
data
03:50
sorry for my bad
03:51
handwriting
03:52
unstructured
03:53
data
03:54
then
03:56
static
03:57
data
03:58
and
03:59
time
04:00
time
04:01
series
04:02
data
04:03
we can see
04:04
this
04:05
that
04:06
you have
04:07
a problem
04:08
that
04:09
i have a problem
04:10
that
04:11
i have a problem
04:12
i have a problem
04:13
then
04:14
existing data
04:15
data
04:16
data
04:17
you have a
04:20
indicate
04:21
what
04:22
you have
04:24
third step
04:25
third step
04:26
third step
04:28
is that you have to define this thing
04:33
that, okay, this is a problem, this is a data,
04:38
so what is my success, defined success?
04:41
The third step is that
04:46
I have to evaluate this on what criteria
04:50
I have to evaluate this.
04:52
Do you understand?
04:54
That I have a problem.
04:56
I have a data, I have a problem, but I think
05:00
the problem is that the evaluation criteria is
05:04
which criteria is called?
05:06
What I can evaluate?
05:08
What I can see in this case?
05:10
For example, I have to look at the road center line.
05:12
This is the criteria.
05:14
If you have to meet this criteria,
05:18
you will say successfully,
05:20
I have to adjust the mirror.
05:22
If you have to meet this criteria,
05:24
then you have successfully solved the problem.
05:30
This criteria is the center line.
05:32
However, this center line
05:34
should meet how much?
05:36
It should be 99.99%?
05:38
Of course, it is not.
05:40
However, it should be so much
05:42
that it should be in frame.
05:44
So basically, you have to define the percentage
05:46
that you have to define
05:48
that 97%
05:50
or let's say,
05:52
let's say,
05:54
80% success will be in this case.
05:56
However, imagine that
05:58
you are making a autopilot
06:00
which will land
06:02
So in that case,
06:04
one degree difference will make our break.
06:06
So in that case,
06:08
80% success will make our break.
06:10
So in that case,
06:12
80% success will make our break.
06:14
So this is the evaluation.
06:16
The third step is to evaluate.
06:18
Please don't mind my English spelling.
06:20
Okay.
06:22
Okay.
06:24
We have three steps.
06:26
We have basically
06:28
the problem of the definition
06:30
that we need to do.
06:32
What do we need to do?
06:34
Okay.
06:36
What do we need to do?
06:38
What do we need to do?
06:40
I have a view.
06:41
The view is on the left side
06:43
or down.
06:45
Okay.
06:46
I will show the data that I have.
06:47
The data is the features.
06:48
Good.
06:49
If I want to adjust the mirror
06:51
to the size of the mirror
06:52
to be successful,
06:53
what do I need to do?
06:54
It is a success criteria.
06:56
It is what the criteria will be
06:57
that the road
06:58
is the point of the center line
06:59
that I need to do
07:00
to cut the mirror
07:02
from this line.
07:03
That is the basic percentage
07:04
in my percentage.
07:05
Excellent.
07:07
Now, let's go to the fourth step
07:09
machine learning
07:10
we have
07:12
idea
07:14
we need
07:15
success criteria
07:17
which we have
07:18
data
07:19
we need
07:20
data
07:21
which we have
07:22
the
07:23
success criteria
07:24
meet
07:25
the efficient way
07:26
which we have
07:27
which we have
07:28
which we have
07:29
which we have
07:30
which we have
07:31
match
07:32
so that we have
07:33
central line
07:34
achieve
07:35
features
07:36
features
07:37
features
07:38
which we have
07:40
model
07:41
okay
07:42
for example
07:43
this case
07:44
first feature
07:45
is
07:46
let's say
07:47
the lights
07:48
look at me
07:49
in my eyes
07:50
this means
07:51
one of the features
07:52
one of the features
07:53
you have to say
07:54
the features
07:55
which we have
07:57
which we have
07:59
directly
08:00
in my eyes
08:01
should not be
08:02
one of the features
08:04
one of the features
08:05
that you have
08:06
which we have
08:08
aspects
08:09
model
08:10
or
08:11
data
08:12
in
08:13
which we have
08:14
pivot points
08:15
for example
08:16
this
08:17
light
08:18
pivot point
08:19
if you have
08:20
more
08:21
than
08:22
the
08:23
range
08:24
range
08:25
will be
08:26
the
08:27
range
08:28
will be
08:29
the
08:30
features
08:31
which you have
08:32
and
08:33
you have
08:34
data
08:35
is
08:36
useful
08:37
and
08:38
the
08:39
fifth step
08:40
is
08:41
basically
08:42
modeling
08:43
in itself
08:44
that you have
08:45
this problem
08:46
for this problem
08:47
this data
08:48
was
08:49
for this
08:50
data
08:51
for this problem
08:52
this solution
08:53
exists
08:54
and
08:55
this
08:56
pivot point
08:57
now
08:58
that
08:59
it
09:00
should
09:01
be
09:02
regression
09:03
or classification
09:04
or clustering
09:05
which
09:06
existing
09:07
model
09:08
which
09:09
will represent
09:10
this problem
09:11
in the solution
09:12
form
09:13
so
09:14
I say
09:15
modeling
09:16
modeling
09:17
modeling
09:18
and
09:19
of course
09:20
machine learning
09:21
is
09:22
experimental
09:23
iterative
09:24
method
09:25
which
09:26
you can
09:27
apply
09:28
then
09:29
you can
09:30
see
09:31
you can
09:32
see
09:33
you can
09:34
see
09:35
you can
09:36
see
09:37
you can
09:38
see
09:39
you can
09:40
see
09:41
slowly
09:42
slowly
09:43
you can
09:44
see
09:45
you can
09:46
see
09:47
this
09:48
process
09:49
is
09:50
complete
09:51
cycle
09:52
this
09:53
process
09:54
let's say
09:56
iteration
09:57
and
10:01
if
10:02
all
10:03
you can
10:04
do
10:05
basically
10:06
we have
10:07
feedback
10:08
loop
10:09
mechanism
10:10
which
10:11
we have
10:12
problem
10:13
define
10:14
this
10:15
level
10:16
and
10:17
when
10:18
we see
10:19
problems
10:20
all
10:21
work
10:22
then
10:23
we have
10:24
to
10:25
evaluate
10:26
that
10:27
regression
10:28
model
10:29
is
10:30
problem
10:31
solve
10:32
but
10:33
it
10:34
doesn't
10:35
solve
10:36
it
10:37
that
10:38
problem
10:39
is
10:40
not
10:41
basically
10:42
it
10:43
is
10:44
automatic
10:45
the
10:46
problem
10:47
was
10:48
very important
10:49
in the
10:50
beginning
10:51
that
10:52
you can
10:53
define
10:54
that
10:55
problem
10:56
is
10:57
very interesting
10:58
if
10:59
you don't
11:00
know
11:01
that
11:02
you can
11:03
do
11:04
is
11:05
hard
11:06
work
11:07
the
11:08
worst
11:09
thing
11:10
that
11:11
you can
11:12
do
11:13
is
11:14
hard work
11:15
the worst thing
11:16
that
11:17
you can
11:18
do
11:19
is
11:20
hard
11:21
work
11:22
so
11:23
this
11:24
framework
11:25
is
11:26
our
11:27
framework
11:28
to
11:29
adjust
11:30
the
11:31
method
11:32
you can
11:33
see
11:34
exactly
11:35
what
11:36
process
11:37
machine
11:38
learning
11:39
applications
11:40
and
11:42
take
11:43
close
11:44
and
11:45
take
11:46
close
11:47
and
11:48
take
11:49
close
11:50
and
11:51
take
11:52
close
Recommended
0:27
|
Up next
Making waterproof shoes with duct tape! πποΈ
MetDaan Creative
5 days ago
0:16
When it's your boyfriend's laundry day! π π€£
MetDaan Creative
5 days ago
0:24
Spreading chocolate cream with a plaster! π π
MetDaan Creative
5 days ago
7:40
Mera Mazloom Hussain | Nadeem Sarwar 2020 | 1442
Gham-e-Hussain | ΨΊΩ Ω ΨΨ³ΫΩΨ
6/9/2025
9:38
Episode 1 β The Curious Boy from Ulm (1879-1895)
Nafees AI Lab
6/8/2025
0:37
Top Fully Funded Scholarships in Germany 2025 π©πͺ | Apply Now!
Nafees AI Lab
6/3/2025
0:25
The Boy and the Dragon β βA Friend in the Skyβ
Nafees AI Lab
6/3/2025
1:03
Why Do Rainbows Appear? π | Physics in Everyday Life
Nafees AI Lab
6/3/2025
0:28
Albert Einstein: The Genius Who Changed Time
Nafees AI Lab
6/1/2025
0:40
Stephen Hawkingβs Life in 60 Seconds: From Wheelchair to the Universe
Nafees AI Lab
6/1/2025
0:36
The Forgotten Night Hitler Was Nearly Killed
Nafees AI Lab
5/29/2025
0:31
Dilip Kumar β The Tragedy King of Bollywood
Nafees AI Lab
5/29/2025
0:45
Marlon Brando β The Rebel Who Redefined Acting | Nafees AI Lab
Nafees AI Lab
5/29/2025
0:29
Charles Darwin β Evolution & Natural Selection | Nafees AI Lab
Nafees AI Lab
5/28/2025
0:30
Isaac Newton β Gravity, Motion & Genius | Nafees AI Lab
Nafees AI Lab
5/28/2025
0:52
Allama Iqbal β Poet of the East | Nafees AI Lab
Nafees AI Lab
5/28/2025
0:53
Marie Curie β Life & Achievements | Nafees AI Lab
Nafees AI Lab
5/28/2025
0:42
Albert Einstein β Life, Legacy & Discoveries | Nafees AI Lab
Nafees AI Lab
5/28/2025
10:08
Pandas Class 10: Manipulating Data β Sorting, Renaming & More
Nafees AI Lab
5/28/2025
12:30
Pandas Class 09: Adding & Removing Data in Pandas
Nafees AI Lab
5/28/2025
11:25
Pandas Class 08: Changing Data in Pandas β Replace & Update
Nafees AI Lab
5/28/2025
14:15
Pandas Class 07: More on Data Selection in Pandas
Nafees AI Lab
5/28/2025
13:58
Pandas Class 06: Data Selection in Pandas β iloc & loc
Nafees AI Lab
5/28/2025
6:59
Pandas Class 05: Describing & Summarizing Data in Pandas
Nafees AI Lab
5/26/2025
7:00
Pandas Class 04 β How to Import CSV, Excel & JSON Files
Nafees AI Lab
5/26/2025
