1 00:00:00,360 --> 00:00:08,460 Hello and welcome to this lecture, where we are going to work with the deep library to solve the problem 2 00:00:08,460 --> 00:00:11,250 about the transport of products. 3 00:00:11,610 --> 00:00:19,950 The good point is that we will be able to compare the results of the library with the results of our 4 00:00:19,950 --> 00:00:22,200 implementation from scratch. 5 00:00:22,500 --> 00:00:31,020 If you want to know more detail about this library, you can follow this link here to get more information. 6 00:00:31,410 --> 00:00:41,190 We can see a brief description of the library and as we can see here, it includes the following features 7 00:00:41,580 --> 00:00:46,410 genetic algorithms, which is the focus of this class. 8 00:00:46,760 --> 00:00:56,760 However, we can see more features, for example, Evolution Strategies mode, the objective optimization 9 00:00:57,060 --> 00:01:02,190 cooperative and the competitive evolution, and many others. 10 00:01:02,490 --> 00:01:12,180 As we saw in the introduction lecture, the genetic algorithms are just a part of the evolutionary algorithms. 11 00:01:12,480 --> 00:01:22,290 So if you want to know more detail about this topic, I recommend that you go deep into these documentation 12 00:01:22,290 --> 00:01:28,590 here to see and to learn more about all other evolutionary algorithms. 13 00:01:28,920 --> 00:01:33,060 Now let's install the library PIP. 14 00:01:33,990 --> 00:01:37,010 Install the App Room. 15 00:01:37,320 --> 00:01:48,270 This sale of codes we can see here that it is downloading the app version one point three point one. 16 00:01:48,540 --> 00:01:52,350 So we are using this version of the library. 17 00:01:52,710 --> 00:01:58,320 In case you have a different version, you can install these. 18 00:01:58,320 --> 00:02:04,080 A specific one should not have problems with different versions. 19 00:02:04,440 --> 00:02:07,590 You can just type PIP install. 20 00:02:07,980 --> 00:02:17,880 The app equals one point three point one, so the codes we are going to implement will be compatible 21 00:02:17,880 --> 00:02:19,950 with this version here. 22 00:02:20,190 --> 00:02:24,930 Now we need to define the class of products. 23 00:02:25,450 --> 00:02:28,380 We'll just paste this code to here. 24 00:02:28,770 --> 00:02:33,150 This code we implement as before and I will just paste. 25 00:02:33,150 --> 00:02:40,350 It should be more organized than we need to create a list of variables. 26 00:02:40,920 --> 00:02:51,360 I will open a new set of codes and based these other codes we implement, as before, the list of products 27 00:02:51,630 --> 00:02:58,080 we are adding in this list before parting products off our case study. 28 00:02:58,530 --> 00:03:04,500 Then we create this basis prices and names lists. 29 00:03:04,980 --> 00:03:15,060 We have bands the values to each one of then this base limit, population size, mutation probability 30 00:03:15,060 --> 00:03:17,310 and number of generations. 31 00:03:17,730 --> 00:03:22,710 Now we run this so of codes to create these variables. 32 00:03:23,220 --> 00:03:25,860 The next step is to eat birds. 33 00:03:26,220 --> 00:03:31,170 The deep library the first one in part no bite. 34 00:03:31,560 --> 00:03:37,830 This is a scientific library in Python to work with matrices and vectors. 35 00:03:38,070 --> 00:03:39,840 We will use it later. 36 00:03:40,290 --> 00:03:54,020 F-words Random to generate random numbers from deep import base from the imports. 37 00:03:54,420 --> 00:04:04,080 Create third from the import algorithms and finally from the deep library. 38 00:04:04,350 --> 00:04:08,220 Let's import the tools package. 39 00:04:08,430 --> 00:04:13,110 We are going to use them in some means run these codes. 40 00:04:13,440 --> 00:04:20,760 No error was returns, so it means that the installation was made successfully. 41 00:04:20,940 --> 00:04:27,690 Now we need to create the fit in this function, which will be exactly the same. 42 00:04:27,780 --> 00:04:29,580 We implemented before. 43 00:04:29,850 --> 00:04:36,900 So when working with libraries, you must implement the fitness function. 44 00:04:37,410 --> 00:04:42,000 Just a reminder that we have this list here. 45 00:04:42,450 --> 00:04:51,150 Zero means that we are not going to load the product on the truck, and one means that we are going 46 00:04:51,150 --> 00:04:53,700 to load the product in the truck. 47 00:04:54,150 --> 00:04:56,100 We can create this function. 48 00:04:56,520 --> 00:04:58,310 Death Fitness. 49 00:04:58,530 --> 00:04:59,850 It will receive. 50 00:04:59,940 --> 00:05:07,750 As a barometer, a solution or a chromosome, now we can create the cost variable. 51 00:05:08,070 --> 00:05:15,510 We are going to return to this variable, which will be equal to this sum of all prizes. 52 00:05:15,760 --> 00:05:22,470 Lest lecturer, we used a different terminology to represent this variable. 53 00:05:22,800 --> 00:05:32,340 But it's important to know that when you work with genetic algorithms, it is common to see these word 54 00:05:32,460 --> 00:05:36,960 cost, which means the results of the fitness function. 55 00:05:38,100 --> 00:05:42,240 Let's create some space equals zero. 56 00:05:42,600 --> 00:05:51,360 Now we need to go through each one of the zeros and ones off the solution. 57 00:05:51,720 --> 00:05:59,880 For I in range length solution, we are going through each one of the zeros and ones. 58 00:06:00,300 --> 00:06:06,630 Let's check if Solution Position II equals one. 59 00:06:07,200 --> 00:06:18,600 Then let's increment the cost plus equal prices position, i.e. we are accessing this list here. 60 00:06:19,080 --> 00:06:22,410 Then we need to add space. 61 00:06:22,800 --> 00:06:28,890 Some space plus equal spaces position i. 62 00:06:29,310 --> 00:06:33,450 It is this list of spaces here. 63 00:06:33,960 --> 00:06:44,970 Then when these short verify after default, look, if some space greater than the limits, the cost 64 00:06:45,150 --> 00:06:46,920 will be equal to one. 65 00:06:47,370 --> 00:06:50,280 And then let's return the cost. 66 00:06:50,610 --> 00:07:00,510 Something different when implement the genetic algorithms using deep library is that we needs to put 67 00:07:00,510 --> 00:07:04,660 this comma here in the next set of calls. 68 00:07:04,680 --> 00:07:07,960 You will understand why we need it. 69 00:07:08,370 --> 00:07:17,250 Let's create the fitness function and now to create some objects related to the deep library. 70 00:07:17,670 --> 00:07:27,300 The first one, let's create the two blocks variable, which will be equal to Bayes dot two blocks. 71 00:07:27,540 --> 00:07:36,690 We need to create two blocks from the base package to initialize the genetic algorithm. 72 00:07:37,080 --> 00:07:39,690 Then let's go creator. 73 00:07:40,020 --> 00:07:42,690 We imported this package here. 74 00:07:43,230 --> 00:07:49,800 Dots create as we have a maximization problem. 75 00:07:50,130 --> 00:07:52,020 We need supports here. 76 00:07:52,800 --> 00:07:58,920 Fit in this max bayes dots fitness. 77 00:07:59,220 --> 00:08:09,780 It is related to the fitness function and now there is another parameter quote weights equals one and 78 00:08:10,050 --> 00:08:11,220 comma here. 79 00:08:11,520 --> 00:08:19,050 See that we have this comma after the value, according should this part here. 80 00:08:19,290 --> 00:08:25,560 When you work with the app, you can return more than one value. 81 00:08:25,680 --> 00:08:35,190 For example, if you had another variable here that you once to optimize your code returns, show variables 82 00:08:35,190 --> 00:08:35,700 here. 83 00:08:36,000 --> 00:08:42,960 For this reason, we need this goal Mahere in this particular case study. 84 00:08:43,200 --> 00:08:53,250 We are returning only the cost, which is the sum of the total price, and this cost is associated with 85 00:08:53,250 --> 00:08:54,390 this value. 86 00:08:54,750 --> 00:09:00,600 One means that the results will be in the range from zero to one. 87 00:09:00,930 --> 00:09:11,580 Then we need to create the individual creator thoughts create and I will type here the name of the glass. 88 00:09:12,000 --> 00:09:14,820 It is related to the deep library. 89 00:09:15,030 --> 00:09:23,550 In the video, we want to generate a list of individuals and now let's type. 90 00:09:23,970 --> 00:09:39,120 Fitness equals creator again thought fitness max, which means that was to maximize the profit or maximize 91 00:09:39,120 --> 00:09:39,930 the cost. 92 00:09:40,350 --> 00:09:43,050 We can run this sale of codes. 93 00:09:43,440 --> 00:09:47,280 Now we needs to register some other objects. 94 00:09:47,610 --> 00:09:59,630 Let's access our toolbox variable, but register now it needs to specify the type of our soldiers. 95 00:09:59,710 --> 00:10:08,170 Washing or that type of the chromosome, let's type A, B, D, R, which means attributes. 96 00:10:08,770 --> 00:10:09,240 Boom. 97 00:10:09,550 --> 00:10:19,630 We are defining a Boolean attributes, which means this solution r d chromosome is composed only of 98 00:10:19,630 --> 00:10:21,580 zeros and ones. 99 00:10:22,090 --> 00:10:28,390 Then we need to specify how we are going to initialize the chromosome. 100 00:10:28,930 --> 00:10:38,440 We will use the transient function as we implement as before, and this solution is composed only of 101 00:10:38,980 --> 00:10:40,750 zeros and ones. 102 00:10:41,170 --> 00:10:50,620 This part here means that this solution will be generated this way here only zeros and ones. 103 00:10:50,860 --> 00:10:54,040 Let's register another component. 104 00:10:54,370 --> 00:11:01,420 The box, but registered the next one is the individual. 105 00:11:01,810 --> 00:11:05,170 Those thoughts it needs. 106 00:11:05,590 --> 00:11:07,000 Repeat it. 107 00:11:07,000 --> 00:11:11,200 The means that we are going to generate random solutions. 108 00:11:11,650 --> 00:11:15,310 Creator but individual. 109 00:11:15,580 --> 00:11:19,120 It is the name of this glass here. 110 00:11:19,510 --> 00:11:32,980 Then we need to specify the box a our bool and this size and equals 14 because there are 14 products 111 00:11:32,980 --> 00:11:34,750 in our case study. 112 00:11:34,930 --> 00:11:44,140 These two lines of code are registering the individuals which are composed of Boolean attributes. 113 00:11:44,590 --> 00:11:48,790 Now we can register the population. 114 00:11:49,150 --> 00:11:52,270 Let's type here, population. 115 00:11:52,570 --> 00:12:03,160 Those dots init repeats to generate a random population, then type the population will be a list. 116 00:12:03,610 --> 00:12:09,730 And now we need to specify the barometer in the video. 117 00:12:09,940 --> 00:12:16,180 So we are creating a population associated with this glass here. 118 00:12:16,540 --> 00:12:28,630 The next step, the box that register evaluates how we are going to evaluate the population using the 119 00:12:28,630 --> 00:12:40,630 fitness function we defined the year, then we need to specify how we are going to apply the crossover. 120 00:12:41,170 --> 00:12:52,030 It is related should the maid's parameter, and then we can access those C x worm lines. 121 00:12:52,420 --> 00:13:00,250 It is the crossover we implement as before, when we have only one cutoff point. 122 00:13:00,550 --> 00:13:06,820 We can see here more options if you was to go deep into this library. 123 00:13:07,090 --> 00:13:17,010 You can see these c x at the beginning means different types of cross over that you can apply. 124 00:13:17,230 --> 00:13:20,620 Let's register another parameter. 125 00:13:21,160 --> 00:13:23,550 It will be the mutation. 126 00:13:24,100 --> 00:13:28,960 Those dots must flip the bit. 127 00:13:29,080 --> 00:13:33,070 It is the mutation we implemented from scratch. 128 00:13:33,370 --> 00:13:42,160 When the number is zero, it is changing the G1 and when the number is one, it is changes to zero. 129 00:13:42,340 --> 00:13:48,280 And the last parameter AI and beep beep beep beep means probability. 130 00:13:48,730 --> 00:13:58,810 We will specify when persons of probability to execute the mutation and the last parameter, the box 131 00:13:59,500 --> 00:14:07,870 thought register how we are going to select the individuals are the best in the videos. 132 00:14:08,560 --> 00:14:15,840 Those dots sale roulette, which is the one we implement as before. 133 00:14:16,150 --> 00:14:23,580 You can see here more options if you want to go deep into genetic algorithms theory. 134 00:14:23,740 --> 00:14:26,350 Let's select the roulette methods. 135 00:14:26,680 --> 00:14:29,320 Now we can read these codes. 136 00:14:29,680 --> 00:14:39,820 There is one error here just correct this typo run again, and the basic structure of the genetic algorithm 137 00:14:39,820 --> 00:14:46,210 is already implemented, and in the next lecture, we are going to finish the implementation. 138 00:14:46,540 --> 00:14:47,230 See you there!