1
00:00:00,990 --> 00:00:07,940
So far, we have been able to change many features of our neural network include the feting network,

2
00:00:08,280 --> 00:00:12,240
we also change the training function, the performance function.

3
00:00:12,700 --> 00:00:15,010
But how about the plot function?

4
00:00:15,630 --> 00:00:17,790
Let me run the program one more time.

5
00:00:19,110 --> 00:00:27,300
As you can see, there are some information that I haven't seen them in this code like Époque time performance

6
00:00:27,300 --> 00:00:28,130
gradient.

7
00:00:28,530 --> 00:00:31,110
How can I change these parameters?

8
00:00:31,600 --> 00:00:32,250
Let's see.

9
00:00:32,520 --> 00:00:34,440
Back to the command window.

10
00:00:37,470 --> 00:00:45,690
Let's type net to see different parameters and features the government has, this is actually a very

11
00:00:45,690 --> 00:00:46,490
long list.

12
00:00:47,070 --> 00:00:54,960
This is a name and some other data, a scroll down until we see airports.

13
00:00:54,960 --> 00:00:58,150
Times go mean wait anymore.

14
00:00:58,560 --> 00:01:01,140
These are in the train Parang.

15
00:01:01,680 --> 00:01:03,240
That's what I'm looking for.

16
00:01:03,240 --> 00:01:09,030
And I want to change type net dot parum.

17
00:01:09,930 --> 00:01:10,560
Oh sure.

18
00:01:10,580 --> 00:01:12,310
It should be twin porro.

19
00:01:12,450 --> 00:01:13,390
Let's back here.

20
00:01:13,410 --> 00:01:14,440
Let me check it again.

21
00:01:16,320 --> 00:01:19,500
Train for the net.

22
00:01:19,710 --> 00:01:21,880
That train.

23
00:01:24,270 --> 00:01:27,930
There are several parameters and functions that we can change here.

24
00:01:28,590 --> 00:01:30,270
The first one is show window.

25
00:01:30,280 --> 00:01:31,590
Show window is true.

26
00:01:32,310 --> 00:01:39,330
When I'm running the program, the window that we can see, it's just that we can just control it and

27
00:01:39,330 --> 00:01:42,140
turn to the fall so we won't see it anymore.

28
00:01:42,450 --> 00:01:47,580
So let me start with show window and see how it can change over program.

29
00:01:48,180 --> 00:01:51,120
I'm going to start typing in here from the net.

30
00:01:51,780 --> 00:01:56,610
I'm looking for trained Parap net.

31
00:01:56,610 --> 00:02:10,920
That train is a big program and then dot this time showed just type it exactly as you see Linda.

32
00:02:12,210 --> 00:02:14,340
I want it to be false.

33
00:02:16,610 --> 00:02:18,820
Let's try it again, OK?

34
00:02:19,310 --> 00:02:23,610
The only window that showed up was the custom neural network.

35
00:02:23,630 --> 00:02:26,420
But the other window can see it here.

36
00:02:27,410 --> 00:02:31,240
Now, let's try different parameters.

37
00:02:32,180 --> 00:02:38,690
If you don't want to see this val performance test performance every time you can come here and make

38
00:02:38,690 --> 00:02:39,710
them as a command.

39
00:02:43,160 --> 00:02:53,330
And performance here and again, OK, I won't see any of this information in my command was the only

40
00:02:53,570 --> 00:02:57,320
thing that happened was my neural network back again here.

41
00:02:57,320 --> 00:03:00,060
Never stopped trying

42
00:03:03,050 --> 00:03:04,850
to check different parameters.

43
00:03:04,850 --> 00:03:08,750
The next one that I want to show is show command line.

44
00:03:09,050 --> 00:03:09,890
This is false.

45
00:03:10,310 --> 00:03:14,090
As you can see, the statue's already changed from true to false.

46
00:03:14,540 --> 00:03:18,770
Next one that I want to change is a show command line typed here.

47
00:03:33,270 --> 00:03:38,340
Next Dot, Twain or DOT.

48
00:03:40,860 --> 00:03:42,540
Show, command line.

49
00:03:46,140 --> 00:03:47,580
Equals to true.

50
00:03:50,510 --> 00:03:53,090
OK, let's run the program one more time.

51
00:03:55,010 --> 00:03:55,600
Very well.

52
00:03:56,240 --> 00:04:03,680
I can see some information here like training depot and a training time and performance degrading,

53
00:04:03,680 --> 00:04:06,080
demeaning in some other information.

54
00:04:07,370 --> 00:04:09,300
Let's try this one again.

55
00:04:10,580 --> 00:04:16,870
It also says that in a command line, it's just going to show the frequency of twenty five.

56
00:04:17,030 --> 00:04:23,100
It means every twenty five times every twenty five, it will show it in a command.

57
00:04:23,630 --> 00:04:27,950
Let me change this one to ten or we can change it to one.

58
00:04:27,950 --> 00:04:38,950
But give me a swordfight 10 again that doesn't train from show to one.

59
00:04:39,620 --> 00:04:45,920
So I should be able to see all the training information is let me run it here.

60
00:04:46,760 --> 00:04:51,740
If you don't want to see this window anymore and then you need to make this one as a command as well,

61
00:04:51,980 --> 00:04:54,280
the unit does make it as a command.

62
00:04:55,520 --> 00:04:59,030
And here we can see all the training.

63
00:04:59,030 --> 00:05:04,250
We can see the Époque time performance gradient, all the information that we were able to see in a

64
00:05:04,490 --> 00:05:07,340
graphical user interface before this time.

65
00:05:07,340 --> 00:05:13,100
We can see that for all the epochs that we had before, here they are.

66
00:05:13,100 --> 00:05:18,410
Twenty eight Époque have actually more from zero is starting from zero.

67
00:05:18,800 --> 00:05:26,130
You can use this information for analyzing your data and you can actually use them in a very good way.

68
00:05:26,150 --> 00:05:31,360
Now go back to the command box and type NetJets Train Prall.

69
00:05:31,670 --> 00:05:33,910
Here is the number of epochs.

70
00:05:34,070 --> 00:05:39,710
It means these next four can train itself up to 1000 times.

71
00:05:40,070 --> 00:05:47,180
If it reached 1000 times and didn't stop by any other parameters, then it stopped training.

72
00:05:47,280 --> 00:05:49,720
It means we won't have any results.

73
00:05:49,730 --> 00:05:52,370
So it's already 1000 times stop the training.

74
00:05:52,730 --> 00:06:00,110
But what if we want to have something less like just tried four, four times and after that, if you

75
00:06:00,110 --> 00:06:02,550
didn't get a new result just to stop the training.

76
00:06:03,050 --> 00:06:05,180
So let me show it to you here.

77
00:06:07,210 --> 00:06:18,310
Next, dot train haraam dot, this time airports equal to five.

78
00:06:19,370 --> 00:06:23,530
Only train this network for five times, and after that it should stopped raining.

79
00:06:25,760 --> 00:06:35,330
Change this one back to true, because I like to see the graphical user interface and this one, I don't

80
00:06:35,330 --> 00:06:38,120
need it back to false.

81
00:06:40,040 --> 00:06:47,300
Tonette Dortch, train product Epagogix five, now let's run the program here we can see the maximum

82
00:06:47,300 --> 00:06:56,300
Époque reached five iteration was the maximum number of times that I let my neural network to train

83
00:06:56,300 --> 00:07:02,030
itself and after five times training Idriz to stop their process of training.

84
00:07:03,330 --> 00:07:08,480
Well, you can see that we can change any of these parameters, the other parameter, and then you can

85
00:07:08,480 --> 00:07:09,470
change it a time.

86
00:07:09,770 --> 00:07:11,110
There is no time here.

87
00:07:11,120 --> 00:07:11,900
Let me show you.

88
00:07:12,730 --> 00:07:17,570
Show it to you in a parameters for the time we have infinity.

89
00:07:18,380 --> 00:07:25,010
If you want to train your network in a certain time, suppose you have many data.

90
00:07:25,010 --> 00:07:27,530
The number of your data is a lot, actually.

91
00:07:27,740 --> 00:07:34,400
The number of samples then you have any you want to limit the time that your P C trains the network.

92
00:07:34,520 --> 00:07:40,070
You want to have it only for three minutes or five minutes, then you can just limited here to use net

93
00:07:40,220 --> 00:07:41,150
train program.

94
00:07:41,150 --> 00:07:47,120
Dot time equals two, for example, five minutes or next one is a goal.

95
00:07:47,450 --> 00:07:48,830
The performance goal.

96
00:07:49,250 --> 00:07:50,480
This is zero.

97
00:07:50,480 --> 00:07:58,550
It means if the network reached zero error, Dennis, stop the training because that's the best scenario

98
00:07:58,670 --> 00:08:00,220
and best situation that we have.

99
00:08:00,530 --> 00:08:08,210
But in reality, we really cannot reach zero, especially if you are having some samples like for the

100
00:08:08,210 --> 00:08:17,140
held parameters for a stress level of the students or some data like that, we can't really reach Zwarte.

101
00:08:18,280 --> 00:08:31,290
Let me put this back to 1000 letters, 100 because five is not enough at all, train all that gold I

102
00:08:31,290 --> 00:08:32,790
can set a goal for.

103
00:08:37,260 --> 00:08:43,320
Zero point, maybe zero zero three, something like that, or let's.

104
00:08:45,690 --> 00:08:54,880
We can see that it has already said the performance to zero point zero zero one where we go, what else

105
00:08:54,900 --> 00:09:05,160
we have here, we have the mean gradient and more information that you can change them based on a data

106
00:09:05,310 --> 00:09:08,820
base and samples that you have and the type of your data.

107
00:09:09,870 --> 00:09:17,120
In the next session, I'm going to discuss Dysport to see you on the next session.