Original subtitles

1

Hello and welcome back to the course on deploring today we're going to wrap up with back propagation.

2

All right so we're you know pretty much everything we need to know about what happens in in your all

3

that we know that there's a process called Forward propagation where information is entered into the

4

input layer and then it's propagated forward to get our white hats our output values and then we compare

5

those to the actual values that we have in our training set.

6

And then we calculate the errors then the errors are back propagated through the network in the opposite

7

direction and that allows us to train the network by adjusting the weights.

8

So the one key important thing to remember here is that back propagation is an advanced algorithm driven

9

by very interesting and sophisticated mathematics which allows us to adjust the weights.

10

All of them at the same time all the weights are adjusted simultaneously.

11

So if we were doing this manually or if we're coming up a very different type of algorithm than Even

12

if we calculated the error and then we were trying to understand what effect each of the weights has

13

on the error we'd have to somehow adjust each of the weights independent independently or individually.

14

The huge advantage of backwardation and it's a key thing to remember is that during the process of back

15

propagation simply because of the way the algorithm is structured.

16

You are able to adjust all the way at the same time so you basically know which part of the error each

17

of your weights in the neural network is responsible for.

18

Now that is the key fundamental underlying principle of back propagation.

19

And this was why it picked up so rapidly in the 1980s and this was a major breakthrough.

20

And if you'd like to learn more about that and how exactly the mathematics works in the background then

21

a good article which we've already mentioned is the neural networks and deep learning is actually a

22

book by Michael Nielsen.

23

You'll find the mathematics written out and it will help you understand how exactly this is possible.

24

But for now for our purposes if from an intuition point of view the important part is to remember that

25

that's what backwardation does.

26

It adjusts all of the weights at the same time.

27

And now we're going to just wrap everything up with a step by step walkthrough of what happens in the

28

training of a neural network.

29

All right so step one we randomly initialized the weights to small numbers close to zero but not zero.

30

We didn't really focus on the initialization of weights during the intuition tutorials but then we have

31

to start somewhere and they are initialized with random values near zero.

32

And from there through the process for propagation by propagation these weights are adjusted until the

33

error is minimized.

34

So the cost function is minimized.

35

Then step two inputs the first observation all your data sets to the first row into the input Lehre

36

each feature is one input.

37

So basically take the combs and put them into the input nodes separately for propagation from left to

38

right.

39

The neurons are activated in a way that they pick in our vision neurons activation is limited by the

40

weights the weights basically determine how important each neurons activation is then propagate the

41

activation until getting the produce a result y hat.

42

In this case.

43

So basically you propagate from left to right.

44

You go all the way until you get to and you get your y hat.

45

Then compare the result to the actual result.

46

Measure the generated error and then you do the backwardation from right to left the air is bipolar

47

again.

48

Update the weights according to how much they are responsible for the error.

49

Again you are able to calculate that because of the way the back propagated perturbation algorithm is

50

structured the learning rate decides by how much we update the weights the learning rate as parameter

51

you can control in your neural network.

52

Step 6 repeat steps 1 2 5 and update the weights after each observation.

53

That is called reinforcement learning and in our case that was stochastic gradient descent or repeat

54

steps 1 to 5.

55

But that way it's only after a batch of observations or batch learning it's either a full gradient descent

56

or badge green Nissan or mini batched gradient descent and step seven when the whole train had passed

57

through artificial neural network that makes an epoch redo more epochs.

58

So basically just keep doing that and doing that and doing that and to allow your neural network to

59

train better and better and better and constantly adjust itself as you minimize the cost function.

60

So there we go.

61

Those are the steps you need to take to build your artificial neural networks and train it.

62

And these are the steps that you will be taking till I've had lunch in the practical tutorials.

63

Wish you the best of luck and I look forward to seeing you next time.

64

Until then enjoy the learning.