1 00:00:00,880 --> 00:00:08,050 Hello and welcome to this lecture, where you were going to learn the intuition and we will implement 2 00:00:08,080 --> 00:00:18,070 another genetic operator, which is mutation, remembering that we are in this part of the image and 3 00:00:18,070 --> 00:00:27,130 after we are going to put all the codes together to build the complete genetic algorithm, mutation 4 00:00:27,430 --> 00:00:33,970 creates diversity by randomly changing the genes of the chromosome. 5 00:00:34,510 --> 00:00:40,900 Here we have the representation of a chromosome or a solution to the problem. 6 00:00:41,410 --> 00:00:48,400 And remember, that's each one and zero is called gene in crossover. 7 00:00:48,580 --> 00:00:57,790 We create a new individual while in mutation, we just randomly change one of the genes. 8 00:00:58,150 --> 00:01:06,610 For example, we can select this gene here and change the value from one to zero. 9 00:01:06,940 --> 00:01:15,700 Or we can select this or other gene here and change the value from zero to one. 10 00:01:16,300 --> 00:01:24,370 These are genetic operator is supplied less often just at its happens in nature. 11 00:01:24,640 --> 00:01:30,040 So crossover is much more frequent than mutation. 12 00:01:30,580 --> 00:01:37,360 We change the genes based on that probability that is usually very small. 13 00:01:37,630 --> 00:01:42,580 For example, one person's two persons are three percent. 14 00:01:43,270 --> 00:01:53,530 Now let's go back to our Google collab file and implement this new genetic operator after a cross over. 15 00:01:54,340 --> 00:02:04,480 We will define a new functional BAFF mutation, and it will receive as a parameter did the full parameter, 16 00:02:04,480 --> 00:02:05,800 which is self. 17 00:02:06,190 --> 00:02:16,780 And also, we will send the here the probability rates, which will indicate the probability of performing 18 00:02:16,810 --> 00:02:17,800 mutation. 19 00:02:18,490 --> 00:02:26,590 How will we implement a far loop for I in range length self? 20 00:02:27,010 --> 00:02:30,460 Let's ex is here the chromosome. 21 00:02:30,790 --> 00:02:35,440 It means that we will go through each one of the genes. 22 00:02:36,010 --> 00:02:46,420 Now we will generate a random number if random, less than the rate than we will. 23 00:02:46,420 --> 00:02:49,150 What BLI mutation, for example. 24 00:02:49,330 --> 00:02:54,310 Let's suppose that's the mutation rate is one person's. 25 00:02:54,670 --> 00:03:05,430 The value would be zero point zero one and then we will generate a random number when we run this code 26 00:03:05,440 --> 00:03:05,950 here. 27 00:03:06,190 --> 00:03:10,740 We will get that number in the range from zero to one. 28 00:03:11,170 --> 00:03:22,150 We can run these codes a couple of times and we will see that the probability of selecting and no less 29 00:03:22,150 --> 00:03:26,380 than zero point zero one is very low. 30 00:03:26,650 --> 00:03:35,560 We can still run in this codes and see that we always get numbers over this value here. 31 00:03:35,860 --> 00:03:39,730 So the probability of mutation is very low. 32 00:03:40,180 --> 00:03:50,320 Now we can implement and if state demands, if self Dot's chromosome, which means these are to report 33 00:03:50,320 --> 00:03:53,560 the year which presents this solution. 34 00:03:54,070 --> 00:04:06,040 If the value of the chromosome or the gene imposition IE equals one, then we need to change this value 35 00:04:06,040 --> 00:04:06,850 to zero. 36 00:04:07,210 --> 00:04:14,190 So self chromosome position ie equals zero. 37 00:04:14,470 --> 00:04:19,210 If the value is one, we change it to zero. 38 00:04:19,580 --> 00:04:26,260 Otherwise, if the value is zero, then we wax as again. 39 00:04:26,260 --> 00:04:33,760 The chromosome in position, i.e. we are accessing the gene equals one. 40 00:04:34,030 --> 00:04:39,310 So if the value is one, the result will be zero. 41 00:04:39,520 --> 00:04:45,910 And if the value of the gene is zero, the result will be one. 42 00:04:46,300 --> 00:04:58,030 Now we can return sales, which means that we will return the object itself because we are changing 43 00:04:58,030 --> 00:05:00,050 all the values here. 44 00:05:00,400 --> 00:05:06,340 You see the results, we can implement a brilliant statement here. 45 00:05:06,700 --> 00:05:15,640 For example, before and let's start here, Self Thoughts chromosome. 46 00:05:15,970 --> 00:05:22,870 And after changing the genes after the for loop, let's wait here after. 47 00:05:23,200 --> 00:05:31,660 And again, self chromosome, let's recreate the individual glass. 48 00:05:31,930 --> 00:05:35,350 And now we will perform some tests. 49 00:05:35,590 --> 00:05:39,460 We are defining the first individual. 50 00:05:40,270 --> 00:05:42,580 This one's in the video. 51 00:05:42,880 --> 00:05:46,090 Here we are, generating the children. 52 00:05:46,610 --> 00:05:52,030 We run this codes again and we can access individual one. 53 00:05:52,570 --> 00:05:58,510 Let's call the object, and now we'll call the mutation. 54 00:05:58,810 --> 00:06:04,540 Let's suppose the probability of mutation is five percent. 55 00:06:04,870 --> 00:06:09,160 So we can type here zero point zero five. 56 00:06:09,550 --> 00:06:11,290 Let's run this codes. 57 00:06:11,680 --> 00:06:17,470 We can see here before mutation and after mutation. 58 00:06:17,890 --> 00:06:20,620 All values are the same. 59 00:06:20,920 --> 00:06:32,560 So it means that when we ran the Randel codes, the values were not lower than zero point zero five. 60 00:06:32,800 --> 00:06:34,840 We can try to run again. 61 00:06:35,350 --> 00:06:41,020 See here that there is one change before this gene. 62 00:06:41,020 --> 00:06:42,850 Here was one. 63 00:06:43,120 --> 00:06:52,270 And after mutation, they value was converted to zero and all other genes keep the same. 64 00:06:52,660 --> 00:06:57,680 We can try to run similar times in this equation. 65 00:06:57,700 --> 00:07:04,180 We can see that's before the value of this gene was zero. 66 00:07:04,420 --> 00:07:09,220 And after applying mutation, this value now is one. 67 00:07:09,730 --> 00:07:15,580 If you want to see more changes, we can put here zero point five. 68 00:07:15,910 --> 00:07:21,220 It means that the probability of mutation is 50 percent. 69 00:07:21,490 --> 00:07:23,920 We can see more changes here. 70 00:07:24,250 --> 00:07:26,170 Zero and one. 71 00:07:26,470 --> 00:07:27,960 One and zero. 72 00:07:27,970 --> 00:07:35,080 And so one, as I told you before, we usually keep a small number here. 73 00:07:35,350 --> 00:07:40,570 IRA commands that you use from one to five percent. 74 00:07:40,900 --> 00:07:48,190 We can try to run again and see the results, and in the next lecture, we are going to continue the 75 00:07:48,190 --> 00:07:49,120 implementation. 76 00:07:49,570 --> 00:07:50,290 See you there.