﻿1
00:00:00,266 --> 00:00:03,933
Hello and welcome to this section
on Convolutional Neural Networks.

2
00:00:04,033 --> 00:00:07,333
Super excited that you're joining us
for this section as well.

3
00:00:07,600 --> 00:00:11,000
And today we're going
to cover off the plan of attack.

4
00:00:11,033 --> 00:00:13,566
How are we going to learn everything
in this section?

5
00:00:13,566 --> 00:00:15,000
There's so much to learn.

6
00:00:15,000 --> 00:00:17,366
let's see
how are we going to approach this.

7
00:00:17,366 --> 00:00:19,500
All right.
What do we learn in this section?

8
00:00:19,500 --> 00:00:23,333
First of all, we'll talk about
what convolutional networks actually are

9
00:00:23,333 --> 00:00:27,066
very important to understand
the end goal that you're working towards

10
00:00:27,066 --> 00:00:28,700
before you actually start
working towards it.

11
00:00:28,700 --> 00:00:30,266
So what about features?

12
00:00:30,266 --> 00:00:32,000
We'll have a look
at a few little examples.

13
00:00:32,000 --> 00:00:35,000
We'll compare the human brain
to artificial neural networks

14
00:00:35,000 --> 00:00:36,633
in terms of image recognition.

15
00:00:36,633 --> 00:00:42,000
So it'll be a fun a light tutorial
to get us started for this whole section.

16
00:00:42,466 --> 00:00:47,333
Then we'll talk about step one diving
straight into it convolution operation.

17
00:00:47,333 --> 00:00:52,300
So this part of the course contains
several steps that we need to go through

18
00:00:52,300 --> 00:00:56,966
in order to build, 
a convolutional neural network.

19
00:00:56,966 --> 00:00:58,966
And that's how these tutorials
are going to be broken up.

20
00:00:58,966 --> 00:01:00,433
So this one is going to be step one.

21
00:01:00,433 --> 00:01:06,400
The convolution operation will learn
everything about feature detectors.

22
00:01:06,400 --> 00:01:08,700
We'll talk about which are also filters.

23
00:01:08,700 --> 00:01:10,900
We'll talk about feature maps.

24
00:01:10,900 --> 00:01:14,466
And you know, how what are the different
parameters there, what they mean.

25
00:01:14,466 --> 00:01:17,400
And have a look
at some visual examples as well.

26
00:01:17,400 --> 00:01:22,633
Then we'll talk about step one, part
B, the ReLU layer or ReLU layer

27
00:01:22,933 --> 00:01:26,600
which is the rectified linear unit.

28
00:01:26,600 --> 00:01:30,966
And we'll talk about 
why linearity is not good

29
00:01:30,966 --> 00:01:36,300
and how we want more non-linearity
in our network for image recognition.

30
00:01:36,733 --> 00:01:38,700
Then we'll talk about step two pooling.

31
00:01:38,700 --> 00:01:41,233
And we'll understand, how pooling works.

32
00:01:41,233 --> 00:01:43,100
We'll talk specifically about max pooling.

33
00:01:43,100 --> 00:01:47,433
And we'll also mention a couple of things
about a mean pooling or some pooling and,

34
00:01:47,900 --> 00:01:51,100
other approaches that you can take
to the process of pooling.

35
00:01:51,900 --> 00:01:55,433
Also in this lecture
we'll have a really cool example.

36
00:01:55,433 --> 00:01:59,633
So there'll be a, very visual interactive
tool that we're going to look at.

37
00:01:59,633 --> 00:02:02,733
So make sure to stick around
to the end of that lecture, because that's

38
00:02:02,733 --> 00:02:06,333
going to add a lot of value
to your learning process.

39
00:02:06,633 --> 00:02:08,700
What we're going to discuss at the end
there.

40
00:02:08,700 --> 00:02:10,600
step three flattening.

41
00:02:10,600 --> 00:02:11,800
So here we will.

42
00:02:11,800 --> 00:02:13,000
It's going to be a quick tutorial

43
00:02:13,000 --> 00:02:16,966
on how to proceed from your pooled layers
to your flattened layer.

44
00:02:17,133 --> 00:02:19,400
And then we're going to talk about
a full connection.

45
00:02:19,400 --> 00:02:24,433
So this is the the very, meaty tutorial
that puts everything together

46
00:02:24,433 --> 00:02:25,933
and puts everything into perspective

47
00:02:25,933 --> 00:02:30,400
and actually shows you
how everything works at the end of the day

48
00:02:30,400 --> 00:02:34,200
and how those final neurons understand
how to classify your image.

49
00:02:34,200 --> 00:02:36,066
Very, very important tutorial.

50
00:02:36,066 --> 00:02:38,833
and hopefully that will,

51
00:02:38,833 --> 00:02:42,033
summarize or kind of,
put everything together for you.

52
00:02:42,366 --> 00:02:45,066
And finally we'll have a summary
which will summarize everything

53
00:02:45,066 --> 00:02:46,400
we've talked about.

54
00:02:46,400 --> 00:02:49,033
And as an extra little feature,
I've included

55
00:02:49,033 --> 00:02:52,033
a tutorial on softmax and cross entropy.

56
00:02:52,100 --> 00:02:55,433
So you don't have to take this tutorial,
but I thought it'd be a great addition,

57
00:02:55,733 --> 00:02:59,300
of knowledge, because these are terms
that you will come across

58
00:02:59,300 --> 00:03:01,900
when dealing
with convolutional neural networks.

59
00:03:01,900 --> 00:03:04,500
So, maybe, maybe take it right away.

60
00:03:04,500 --> 00:03:08,066
Maybe, when you come across these terms,
you can you will always know

61
00:03:08,066 --> 00:03:11,066
you can come back to this course and take,

62
00:03:11,066 --> 00:03:14,266
this tutorial to understand
better what softmax and cross-entropy are.

63
00:03:14,600 --> 00:03:18,866
And also, as always throughout,
these, tutorials, there will be

64
00:03:18,866 --> 00:03:22,666
lots of recommended reading for you
to further upskill and get more knowledge.

65
00:03:23,233 --> 00:03:25,933
And on that note, I can't
wait to see you on the first tutorial!

66
00:03:25,933 --> 00:03:27,233
This is going to be a very fun

67
00:03:27,233 --> 00:03:31,200
and exciting section,
and until next time, enjoy deep learning!
