Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated: 1 00:00:00,510 --> 00:00:06,330 Hello and welcome to this new tutorial and this toile we're going to create the brain of our discriminator 2 00:00:06,390 --> 00:00:07,740 now that we have the class. 3 00:00:07,770 --> 00:00:13,590 But the thing is I might break the record for the fastest to toil in the world because actually we already 4 00:00:13,590 --> 00:00:14,670 have everything. 5 00:00:14,760 --> 00:00:20,490 There is nothing more easy to create the brain of the discriminator because basically that's exactly 6 00:00:20,490 --> 00:00:24,540 the same as what we did for the brain of the generator. 7 00:00:24,540 --> 00:00:26,690 We simply need to choose a new name. 8 00:00:26,730 --> 00:00:28,810 We're going to call it net D. 9 00:00:28,890 --> 00:00:35,250 Obviously we need to create an object of the class corresponding to the discriminator that contains 10 00:00:35,250 --> 00:00:39,060 the architecture of the neural network and the void function. 11 00:00:39,060 --> 00:00:40,860 And then that's the same here. 12 00:00:40,890 --> 00:00:47,430 We're going to take our discriminator object and apply the weights in a function that will initialize 13 00:00:47,430 --> 00:00:49,190 the weights the proper way. 14 00:00:49,560 --> 00:00:54,000 And here we are done with the creation of the brain of the discriminator. 15 00:00:54,120 --> 00:00:56,500 So now huge congratulations. 16 00:00:56,520 --> 00:00:57,940 You created two brains. 17 00:00:58,050 --> 00:01:04,650 It's now time to train them make them smart and actually make them artists because they will actually 18 00:01:04,650 --> 00:01:08,110 generate some art by creating some fake images. 19 00:01:08,280 --> 00:01:12,520 But you're going to see that these are actually going to be some beautiful images. 20 00:01:12,540 --> 00:01:15,160 I mean some images that look like something. 21 00:01:15,180 --> 00:01:16,910 So again I have to insist. 22 00:01:16,920 --> 00:01:24,150 Congratulations because not only have created two brains but also you are done with the first huge step 23 00:01:24,270 --> 00:01:25,720 of this implementation. 24 00:01:25,890 --> 00:01:32,160 If we want to divide this implementation into two big parts Well I would say that the first big part 25 00:01:32,310 --> 00:01:34,080 is the creation of these two brains. 26 00:01:34,140 --> 00:01:37,700 And the second part is the training of these two brains. 27 00:01:37,710 --> 00:01:41,340 So you're done with the first big part because you have created these two brains. 28 00:01:41,340 --> 00:01:45,040 And now we're going to move on to the second part to train these two brains. 29 00:01:45,060 --> 00:01:50,220 And of course train them according to the process of degenerative adversarial networks and let's give 30 00:01:50,220 --> 00:01:54,840 a quick reminder of what this process is about this process is two big steps. 31 00:01:54,840 --> 00:02:02,040 The first step is to update the weight of the discriminator neural network and that consists of maximizing 32 00:02:02,040 --> 00:02:04,510 the error calculated by the B.C.E.. 33 00:02:04,530 --> 00:02:05,420 You will see that. 34 00:02:05,520 --> 00:02:11,910 And the second big part and then the second big step of this process is to have data weights of this 35 00:02:11,910 --> 00:02:19,370 time the generator which again will consist of maximizing the area calculated by the bacillus. 36 00:02:19,370 --> 00:02:25,010 All right so if you accept take the challenge to train this brain will join me for the second part of 37 00:02:25,010 --> 00:02:31,430 this implementation and so I'll you very soon to make our decisions the greatest artist of all the computer 38 00:02:31,430 --> 00:02:32,420 vision models. 39 00:02:32,630 --> 00:02:34,460 Until then enjoy computer vision. 4146