1
00:00:00,466 --> 00:00:02,866
Hello and welcome back to the course
on Deep Learning.

2
00:00:02,866 --> 00:00:06,233
I hope you're tracking along
with these intuition tutorials just fine,

3
00:00:06,500 --> 00:00:09,966
and that you had a chance to play around
with everything we've learned so far.

4
00:00:10,500 --> 00:00:12,000
And today we're talking about flattening.

5
00:00:12,000 --> 00:00:14,800
And the good news is that
this is a very simple step.

6
00:00:14,800 --> 00:00:17,800
And this tutorial
is going to be very quick.

7
00:00:17,833 --> 00:00:21,200
And then we'll be able to move on
to the next interesting things.

8
00:00:21,766 --> 00:00:22,100
All right.

9
00:00:22,100 --> 00:00:24,266
So we so far we've got the pulled layer.

10
00:00:24,266 --> 00:00:25,400
I've pulled feature map.

11
00:00:25,400 --> 00:00:30,300
And that is after we apply 
the convolution operation to our image.

12
00:00:30,300 --> 00:00:31,166
And then we apply

13
00:00:31,166 --> 00:00:34,766
pooling to the result of the convolution
which is the convolved image.

14
00:00:35,000 --> 00:00:37,433
And so what are we going to do
with this pooled feature map.

15
00:00:37,433 --> 00:00:41,033
Well we're going to take it and
we're going to flatten it into a column.

16
00:00:41,566 --> 00:00:42,033
so basically

17
00:00:42,033 --> 00:00:46,166
just take the numbers row by row
and put them into this one long column.

18
00:00:47,066 --> 00:00:51,000
And the reason for that is
because we want to later input this into

19
00:00:51,000 --> 00:00:54,633
an artificial neural network
for further processing.

20
00:00:55,166 --> 00:00:58,600
So this is what it looks like
when you have many pooling layers.

21
00:00:58,600 --> 00:01:02,333
Or you have the pooling layers
with many, pooled feature maps,

22
00:01:02,700 --> 00:01:04,133
and then you flatten them.

23
00:01:04,133 --> 00:01:07,166
So you put them into this one long column

24
00:01:07,166 --> 00:01:10,500
sequentially
one after the other, and you get one huge,

25
00:01:11,833 --> 00:01:12,200
vector

26
00:01:12,200 --> 00:01:15,200
of inputs
for an artificial neural network.

27
00:01:15,300 --> 00:01:18,766
And so to sum all of this up,
we've got an input image.

28
00:01:19,200 --> 00:01:21,000
We apply a convolution layer.

29
00:01:21,000 --> 00:01:23,033
And let's not forget

30
00:01:23,033 --> 00:01:25,800
ReLU or rectified rectified

31
00:01:25,800 --> 00:01:29,566
linear units function that we apply
after the convolution layer as well.

32
00:01:29,933 --> 00:01:33,533
And then we apply pooling
and then we flatten everything

33
00:01:33,766 --> 00:01:36,566
into

34
00:01:36,566 --> 00:01:39,833
a long vector
which will be our input layer

35
00:01:40,033 --> 00:01:42,900
for an artificial neural network.

36
00:01:42,900 --> 00:01:44,333
And exactly how that works.

37
00:01:44,333 --> 00:01:47,100
We'll find out in the next tutorial.

38
00:01:47,100 --> 00:01:49,833
Hope you enjoyed today's session and I
look forward to seeing you next time.

39
00:01:49,833 --> 00:01:52,833
Until then, enjoy deep learning.