1
00:00:00,730 --> 00:00:08,410
Hello and welcome to this lecture, where you are going to understand the how to select the best individuals

2
00:00:08,710 --> 00:00:11,320
to create the new generation.

3
00:00:11,680 --> 00:00:20,920
This part here where I am blind in the mouth, as you can see in this image, we first create the initial

4
00:00:20,920 --> 00:00:23,890
population and then evaluate.

5
00:00:23,890 --> 00:00:32,720
It's just a reminder that the evaluation is this some of the products prices we have already learned

6
00:00:32,980 --> 00:00:38,980
how to implement the genetic or braiders crossover and mutation.

7
00:00:39,280 --> 00:00:44,740
However, see, that's before we apply the operators.

8
00:00:45,160 --> 00:00:54,910
We need to select the parents here who are some of the best individuals in the population, as we saw

9
00:00:54,910 --> 00:00:56,260
in the implementation.

10
00:00:56,530 --> 00:01:05,800
Some individuals to have this car are the total prize equal to one, which indicates that the products

11
00:01:05,800 --> 00:01:07,030
exceeded.

12
00:01:07,060 --> 00:01:09,490
The thought, though, is based off the truck.

13
00:01:09,970 --> 00:01:18,940
So these individuals should not be selected because we would generate new bad solutions.

14
00:01:19,360 --> 00:01:28,450
For this reason, we need to select some of the best in the videos to create the next generation.

15
00:01:28,900 --> 00:01:38,710
As a result, the best individuals, those who have the highest prices, will be selected more often

16
00:01:39,010 --> 00:01:42,580
than individuals who have a low price.

17
00:01:43,240 --> 00:01:50,440
The features of the past in the individuals must predominate in the new population.

18
00:01:50,830 --> 00:02:00,670
In short, we will prioritize solutions with higher values as we will dance to generate even better

19
00:02:00,670 --> 00:02:03,940
individuals in the next generations.

20
00:02:04,690 --> 00:02:06,580
Should select individuals?

21
00:02:06,820 --> 00:02:16,450
We are going to use one of the simplest ways which scalds the rule lets methods, as you can see in

22
00:02:16,450 --> 00:02:20,400
this image, to easily understand this method.

23
00:02:20,710 --> 00:02:31,840
Let's consider that each slice is a solution, and let's suppose that the value of solution ASTRI three

24
00:02:31,840 --> 00:02:38,440
here is 1000 and the value of solution asks.

25
00:02:38,470 --> 00:02:47,750
Sure, here is five 5000 remembering that's our goal is to select the highest price.

26
00:02:48,430 --> 00:03:00,220
If you spin the wheel to randomly select a solution, the probability of selecting Eschew is much greater

27
00:03:00,220 --> 00:03:08,770
than select the S3, as this area here is larger than this area here.

28
00:03:09,370 --> 00:03:19,000
However, there is a small probability of select the SE three, and this is very important because even

29
00:03:19,000 --> 00:03:28,000
though we selected the best individuals most of the time, we also need to leave small probabilities

30
00:03:28,330 --> 00:03:30,890
for the individual those that are not.

31
00:03:30,910 --> 00:03:36,520
That's good because it will create some more diversity in the algorithm.

32
00:03:36,910 --> 00:03:46,840
If we always use the best solutions all the time, the population will dance to be composed of increasingly

33
00:03:46,840 --> 00:03:51,130
similar individuals, and we lack diversity.

34
00:03:51,520 --> 00:04:00,970
If you run the algorithm many times, we will almost always have very similar results in the implementation.

35
00:04:01,180 --> 00:04:05,660
We had some solutions that the value was equal to one.

36
00:04:06,100 --> 00:04:14,620
Then the videos that this base occupied by the products is greater than the available space on the truck.

37
00:04:15,100 --> 00:04:19,210
Those individuals will also be here.

38
00:04:19,480 --> 00:04:27,610
However, this slice will be much smaller and there will hardly be selected.

39
00:04:28,030 --> 00:04:36,760
And to implement this methods, we are going to use the function we implement as before where we got

40
00:04:36,760 --> 00:04:39,160
this, some of the evaluations.

41
00:04:39,520 --> 00:04:47,500
So then you can select the individuals and then the next the lecturer we will we implemented step by

42
00:04:47,500 --> 00:04:48,160
step.

43
00:04:48,250 --> 00:04:50,950
So it will be easier to understand.

44
00:04:51,430 --> 00:04:52,150
So there.